South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling Phase 3 studies
Prepared for Trans-Tasman Resources Ltd
October 2013
Authors/Contributors: Mark Hadfield
For any information regarding this report please contact: Neville Ching Contracts Manager Wellington +64-4-856 0597 [email protected]
National Institute of Water & Atmospheric Research Ltd 301 Evans Bay Parade, Greta Point Wellington 6021 Private Bag 14901, Kilbirnie Wellington 6241 New Zealand
Phone +64-4-386 0300 Fax +64-4-386 0574
NIWA Client Report No: WLG2013-36 Report date: October 2013 NIWA Project: TTR11301
© All rights reserved. This publication may not be reproduced or copied in any form without the permission of the copyright owner(s). Such permission is only to be given in accordance with the terms of the client’s contract with NIWA. This copyright extends to all forms of copying and any storage of material in any kind of information retrieval system.
Whilst NIWA has used all reasonable endeavours to ensure that the information contained in this document is accurate, NIWA does not give any express or implied warranty as to the completeness of the information contained herein, or that it will be suitable for any purpose(s) other than those specifically contemplated during the Project or agreed by NIWA and the Client.
Contents
Executive summary ...... 7
1 Introduction ...... 9 1.1 Model setup ...... 10 1.2 Nested grids ...... 10 1.3 Cook Strait model ...... 11 1.4 South Taranaki Bight model ...... 14 1.5 Patea Shoals model ...... 15 1.6 Sediment model setup ...... 15 1.7 River inputs ...... 18
2 Hydrodynamic model evaluation ...... 20 2.1 Field measurements ...... 20 2.2 Tidal current comparison ...... 20 2.3 Sub-tidal current comparison ...... 23
3 Sediment properties and release parameters ...... 28 3.1 Natural sediments ...... 28 3.2 Mining-derived sediments ...... 29
4 Sediment model evaluation ...... 35 4.1 Surface SSC comparison with remote-sensed data ...... 35 4.2 Surface SSC comparison with in situ measurements ...... 35 4.3 Near-bottom SSC comparison with in situ ABS data ...... 39
5 Sediment plume results ...... 42 5.1 Statistical measures of SSC ...... 42 5.2 Suspended sediment source A ...... 46 5.3 Suspended sediment source B ...... 57 5.4 Suspended sediment source A: recovery simulation ...... 64 5.5 Patch source ...... 67 5.6 Freshwater source ...... 70
6 Acknowledgements ...... 71
7 Glossary of abbreviations and terms ...... 72
8 References ...... 73
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Appendix A Model setup details ...... 76 The ROMS vertical grid ...... 76
Appendix B Composition and properties of sediments ...... 77 Size classes ...... 77
Appendix C Model-measurement comparisons ...... 78 Tidal velocity comparison ...... 78 Sub-tidal velocity comparison ...... 81
Tables Table 1-1: Rivers represented in the inner model, with mean freshwater and sediment input rates from WRENZ. 18 Table 2-1: ADCP data availability from the field measurements. 20 Table 3-1: Natural sediment classes and properties. 28 Table 3-2: Hydro-cyclone overflow discharge rates. 30 Table 3-3: De-ored sand discharge rates. 31 Table 3-4: Sediment classes, properties and discharge rates for the ironsand extraction suspended source. 32 Table 3-5: Source locations. 32 Table 3-6: Sediment classes and composition for the deposited patch source. 33 Table 3-7: Sediment classes and composition for the seabed surrounding the deposited patch source. 34
Table A-1: Inner model layer thicknesses. 76 Table B-1: Sediment grain size classification. 77 Table C-1: Comparison of Mtidal ellipse parameters. 80 Table C-2: Mid-depth sub-tidal velocity comparison. 84 Table C-3: Near-surface subtidal velocity comparison. 85 Table C-4: Near-bottom subtidal velocity comparison. 86
Figures Figure 1-1: The Patea Shoals. 9 Figure 1-2: ROMS grids. 11 Figure 1-3: Surface circulation around New Zealand. 12 Figure 1-4: Time-averaged, depth-average velocity from the Cook Strait model. 14 Figure 1-5: Time-averaged, depth-average velocity from the South Taranaki Bight model. 14 Figure 1-6: Bottom boundary layer & sediment bed time series at instrument site 7. 17 Figure 1-7: Daily average SSC vs flow for Whanganui River. 19
Figure 2-1: M2 tidal velocity comparison (ADCP Site 7 Deployment 2). 21
Figure 2-2: M2 tidal current profile comparison (ADCP Site 7 Deployment 2). 22
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Figure 2-3: Tidal velocity comparison for S2, N2 and K1 (ADCP Site 7 Deployment 2). 23 Figure 2-4: Sub-tidal velocity comparison (ADCP Site 7 Deployment 2). 25 Figure 2-5: Comparison of detiding methods (ADCP Site 7 Deployment 2). 26 Figure 4-1: Mean surface concentration of natural sediment. 35 Figure 4-2: SSC time series at near-shore site 11 (Whanganui). 37 Figure 4-3: SSC time series at near-shore site 12 (Kai Iwi). 37 Figure 4-4: SSC time series at near-shore site 13 (Waitotara River). 37 Figure 4-5: SSC time series at near-shore site 14 (Patea). 38 Figure 4-6: SSC time series at near-shore site 15 (Manawapou). 38 Figure 4-7: SSC time series at near-shore site 16 (Ohawe). 38 Figure 4-8: Time series of near-bottom sand concentration at instrument site 6. 40 Figure 4-9: Time series of near-bottom sand concentration at instrument site 7. 40 Figure 4-10: Time series of near-bottom sand concentration at instrument site 8. 40 Figure 4-11: Time series of near-bottom sand concentration at instrument site 10. 41 Figure 5-1: Surface SSC time series at near-shore optics survey site 11 (Whanganui) with derived statistical parameters. 44 Figure 5-2: Surface SSC time series at a site 8 km from mining source A. 45 Figure 5-3: Median near-surface concentration of suspended sediment from mining source A. 47 Figure 5-4: 99th percentile near-surface concentration of suspended sediment from mining source A. 49 Figure 5-5: Median near-bottom concentration of suspended sediment from mining source A. 50 Figure 5-6: 99th percentile near-bottom concentration of suspended sediment from mining source A. 51 Figure 5-7: Maximum 5-day increment in sediment bed thickness for suspended sediment from mining source A. 53 Figure 5-8: Maximum 365-day increment in sediment bed thickness for suspended sediment from mining source A. 54 Figure 5-9: Vertical profiles of SSC statistics for near-shore optics survey site 11 (Whanganui). 56 Figure 5-10: Vertical profiles of SSC statistics for at a site 8 km from mining source A. 56 Figure 5-11: Median near-surface concentration of suspended sediment from mining source B. 58 Figure 5-12: 99th percentile near-surface concentration of suspended sediment from mining source B. 59 Figure 5-13: Median near-bottom concentration of suspended sediment from mining source B. 60 Figure 5-14: 99th percentile near-bottom concentration of suspended sediment from mining source B. 61 Figure 5-15: Maximum 5-day increment in sediment bed thickness for suspended sediment from mining source B. 62 Figure 5-16: Maximum 365-day increment in sediment bed thickness for suspended sediment from mining source B. 63 Figure 5-17: Thickness of mining-derived sediment for the recovery simulation. 65 Figure 5-18: 99th percentile of near-bottom SSC for the recovery simulation. 66 Figure 5-19: Thickness of sediment from the patch source. 68
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 5
Figure 5-20: 99th percentile of near-bottom patch SSC. 69 Figure 5-21: Statistics of surface freshwater concentration from the freshwater source. 70
Figure A-1: ROMS vertical grid schematic. 76 Figure C-1: Tidal ellipse comparison (all ADCP deployments). 78 Figure C-2: Tidal profile comparison (all ADCP deployments). 79 Figure C-3: Sub-tidal variance ellipse comparison (all ADCP deployments). 81 Figure C-4: Sub-tidal along-axis velocity comparison (all ADCP deployments). 82 Figure C-5: Sub-tidal across-axis velocity comparison (all ADCP deployments). 83
Reviewed by Approved for release by
Dr Graham Rickard Dr Andrew Laing
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Executive summary Trans-Tasman Resources Ltd (TTR) plans to extract titanomagnetite sand (ironsand) from an area in South Taranaki Bight. NIWA was commissioned by TTR to investigate potential environmental impacts of the proposed extraction operation. The operation will result in suspended-sediment plumes and sediment deposition on the seabed that may cause environmental effects on the ocean environment. Dispersed sediment plumes could also reach the near-shore coastal environment.
NIWA has undertaken to estimate the concentrations and deposition rates of sediments released from the ironsand extraction operation. The project is completed with the delivery of this report. It is based on the project as defined in a plan provided by TTR on 17 July 2013. It supersedes earlier reports (Hadfield 2011, 2013) based on earlier, now superseded, project plans.
A set of nested model grids has been set up for the ROMS ocean model, an outer grid covering Cook Strait at a resolution of 2 km and a pair of alternative inner grids, one covering South Taranaki Bight and the other a smaller area on Patea Shoals, both at 1 km resolution with the option of refining them to 500 m if necessary.
The modelling system was first used to simulate currents for a 1000-day period. Tidal and sub-tidal modelled currents are compared with measured currents from a total of 10 instrument deployments at 5 sites. For tidal currents, agreement is very good. For sub-tidal currents, agreement is good, and comparable with what has been achieved in other similar modelling studies. The model tends to under-predict variability in currents at the two sites closest to the shore. This is likely to be due to biases in the surface wind data. It may cause the model to underpredict the rate at which river plumes are moved along the coast, but this should not be a serious bias.
The model was then used to simulate natural sediment processes in the area. Comparisons with remote-sensed data and in situ time series showed that the model reproduces observed suspended sediment concentrations (SSCs) reasonably well. Comparison with acoustic backscatter estimates of near-bottom suspended sand concentrations showed that the model over-predicts the rate of sand resuspension at some locations and under-predicts it at others.
The output of sediment (or in one case, freshwater) from the ironsand extraction operation was represented with three sources:
. The suspended source, representing fine sediment (grain size < 90 µm) introduced into suspension via two discharge streams: the overflow from the hydro-cyclone dewatering system and the de-ored sand discharge.
. The patch source, an area of 3 × 2 km representing one year’s discharge of de- ored sand, including a small amount of fine material (grain size 38–90 µm).
. The freshwater source, representing fresh water used to rinse the ironsand concentrate and discharged from hyperbaric filters.
Of these three sources, the suspended source has the greatest impact in terms of the extent and magnitude of the concentrations in the sediment plume.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 7
Two source locations are considered, at the inner end (A) (see Figure 6-3) and the outer end (B) (see Figure 6-11) of the project area.
The suspended source was introduced in a simulation of 1000 days on the South Taranaki Bight domain, with the source operating for 800 days (with 20% down-time) and statistics calculated over the final two years. The analysis of SSC focussed on the median and 99th percentile, comparing values for natural sediments, mining-derived sediments and the combination of the two. There are conceptual complications in carrying out these comparisons, owing to the effects of the sediments on the flow and the interactions between different sediments. These issues are investigated in terms of time series of surface SSC at two sites: a near-shore site near the Whanganui River and an off-shore site 8 km from source location A. At the former site, natural sediments dominate over mining-derived sediments; at the latter site, the mining-derived SSCs were larger than the natural SSCs
Plumes from the suspended source extend to the east-southeast from both source locations, reaching the coast between Patea and Whanganui and with a long tail of low concentrations following the coast towards Kapiti. For source A the highest surface concentrations occur at the source location and are 2.5–5 mg/L (median) and 10-20 mg/L (99th percentile). For source B the plume is located further offshore and the concentrations somewhat lower. In both cases the plume of mining-derived sediment contributes significantly to the total SSC within a few kilometres of the source but is insignificant relative to the natural SSCs near the coast.
Deposition from the suspended source was characterised by two statistics, the maximum 5- day deposition (i.e. the maximum amount of material accumulated over any 5-day interval) and the maximum 365-day deposition. As with SSC, the deposition footprint of the mining- derived sediments can be distinguished from the natural background in the vicinity of the source, but not near the coast.
A “recovery” simulation was set up to investigate the sequence of events when the suspended source is turned off. After 800 days of operation (the beginning of the recovery phase) there is an extensive patch of deposited mining-derived sediment to a thickness of up to 10 mm near the source and around 2–3 mm in a secondary maximum between the Whanganui and Manawatu Rivers. Over the following two years of recovery this patch is eroded away and the associated SSCs reduce considerably.
With the patch source, sands move away from the initial patch location for a distance of up to 10 km over two years. There is no extensive plume of suspended sediment. For various reasons the patch source simulation may underestimate the extent to which patch material is eroded and transported, but even allowing for this bias no extensive plume is expected.
Concentrations of freshwater in the plume from the freshwater source are low, within the 99th percentile concentration less than 0.1%.
The ROMS sediment model predicts the suspended sediment concentration, i.e. the mass of sediment per unit volume, normally expressed in mg/L. However many of the potential effects of suspended sediment involve the interaction of the sediment particles with light. These effects include the visual appearance of sediment from above the water, reduction of underwater visibility and reduction of underwater light levels. These aspects are assessed in a separate report.
8 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
1 Introduction Trans-Tasman Resources Ltd (TTR1) plans to extract titanomagnetite sand (ironsand) from an area in South Taranaki Bight. The plan defines a project area in a roughly triangular region at 20–40 m depth off the South Taranaki coast near Patea. On LINZ chart NZ45 (Figure 1-1) parts of this area are labelled Whenuakura Spur, Graham Bank, Patea Banks and The Rolling Ground. Here the area as a whole will be called the Patea Shoals.
Figure 1-1: The Patea Shoals. Excerpt from LINZ chart NZ45 (http://charts.linz.govt.nz/tifs/nz45.tif).
Under Work Schedules C (Sediment Plume Modelling Phase 2) and M (Sediment Plume Modelling Phase 3) of a contract between NIWA and TTR, NIWA has undertaken to “Estimate the concentrations and deposition rates of sediments released from the sand mining activities, including dredging loading and dumping”. Phase 3 is complete with the delivery of the present report. It is based on the project as defined in a plan provided by TTR on 17 July 2013. It supersedes earlier reports (Hadfield 2011, 2013) based on earlier, now superseded, project plans.
The results in this report are all expressed in terms of sediment mass, either suspended sediment concentration (SSC) in mg/L or deposited mass in kg/m2, the latter normally converted to an equivalent thickness in mm.. However many of the potential effects of suspended sediment involve the interaction of the sediment particles with light. These effects include the visual appearance of sediment from above the water (which involves backscattering of light by particles near the surface), reduction of underwater visibility (which
1 http://www.ttrl.co.nz/.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 9
is related to a quantity called light beam attenuation), and reduction of underwater light levels (which involves light absorption by sediment). The optical effects are not related in a straightforward way to the SSC, because they depend on the size, colour and number of the suspended sediment grains, as well as their mass.
The optical effects of the suspended sediments (natural and mining-derived) is assessed in a separate report (GAll et al. 2013) using an optics model, which requires information on the so-called inherent optical properties (IOP) of the sediment material. IOPs of natural sediment and of material from the TTR pilot ironsand processing plant have been measured by NIWA and has been used in the optical effects modelling.
1.1 Model setup
1.2 Nested grids Sediment plume behaviour was predicted using a modelling system comprising a set of nested domains. The outer domain covered Greater Cook Strait. Two different inner domains were defined and used in different model runs, a larger one covering South Taranaki Bight and another covering a smaller area over Patea Shoals. The model was ROMS (Haidvogel et al. 2008), a widely accepted ocean/coastal model with optional embedded models of suspended-sediment and sediment-bed processes (Warner et al. 2008). In all cases the inner domain is the one on which sediment plumes are simulated.
In the model nesting procedure, the outer model provides time-varying lateral boundary data (temperature, salinity, velocity, sea-surface height) for the inner model. The simulations are carried out separately, with output fields from the outer model saved to files and later post- processed to provide boundary-data files on the inner grid. This process is called one-way, off-line nesting.
The model grids are shown in Figure 1-2.The outer grid covers Greater Cook Strait at 2 km resolution; the larger of the two inner grids covers South Taranaki Bight from Cape Egmont to just north of Kapiti Island; the smaller inner grid extends over Patea Shoals from mid-way between Hawera and Opunake to a few kilometres southeast of Wainui Beach. The inner grids have been implemented at two different resolutions: 1 km and 500 m. The model runs described in this report were carried out on the 1 km grids, with the 500 m grids used in the past to investigate the sensitivity of the model results to the grid resolution.
The bathymetry for the model grids was constructed using several different datasets, combined and gridded with the GMT mapping tools2. The primary dataset was a bathymetry of South Taranaki Bight on a 100 m grid, generated by NIWA (A. Pallentin and R. Gorman, pers. comm. 13 June 2013) and incorporating data from Patea Shoals surveys conducted by TTR and NIWA. This dataset was supplemented with several lower-resolution regional datasets, namely coastline data, land elevation data, continental shelf bathymetric contour data and the GEBCO3 gridded ocean bathymetry.
2 http://gmt.soest.hawaii.edu/ 3 http://www.gebco.net/
10 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
1.3 Cook Strait model The Cook Strait model requires lateral boundary data to generate a realistic flow from west to east through Cook Strait, with the inflowing water having realistic temperature and salinity. The east-west flow is called the D’Urville Current (“DC” in Figure 1-3) and is a robust feature of ocean models. It is driven by the difference in density (and consequently mean sea level) between Tasman Sea and Pacific Ocean waters.
Figure 1-2: ROMS grids. a) Outer and inner grids; b) South Taranaki Bight grid with bathymetry (coloured surface), coastline (yellow), 12 nautical mile (22.2 km) territorial limit (thin white line), project area (thick white line), ADCP sites (dark blue), river locations (blue) and towns (black).
For the present work two different sources of boundary data were tested. One was an application of ROMS to the New Zealand EEZ. The other was a global ocean analysis and prediction system operated by the US Naval Research Laboratory, using the HYCOM4 ocean model. (The specific dataset used here is called Glba08.) The HYCOM system provides daily snapshots of the three-dimensional state of the global ocean on a 1/12° grid; at NIWA we have archived a subset of this data around New Zealand since 2003. The tests indicated that
4 http://hycom.org/
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 11
estimates of currents, plume dispersion and transport in Cook Strait were not sensitive to the source of boundary data. All the simulations described in this model use HYCOM.
Figure 1-3: Surface circulation around New Zealand. Excerpt from “Ocean Circulation New Zealand” (Carter et al. 1998).
Model times are expressed in days relative to a reference time of 00:00 UTC on 1 January 2005. The model was initialised at 2000 days and run to 3000 days. Analyses of statistical quantities (means, percentiles) in the remainder of this report are generally based on the last 730 days of the simulation (21 March 2011 to 20 March 2013), allowing the first 270 days of the simulation for the model to approach equilibrium. (However, note that different aspects of the model approach equilibrium at different rates. Currents adjust within days to weeks, and temperature and salinity within months, but deposited sediments move slowly through the system and do not reach equilibrium within a period of several years.)
In both models the heat flux through the sea surface was calculated using data (6-hourly averages) from a global atmospheric analysis system called the NCEP Reanalysis (Kalnay et al. 1996), with a heat flux correction term that causes the model sea surface temperature (SST) to be nudged towards observed SST (the NOAA Optimum Interpolation 1/4° daily SST dataset, Reynolds et al. 2007). The heat flux correction prevents the modelled SST from departing too far from reality due to any biases in the surface fluxes, but has a negligible effect on day-to-day variability. The surface stress was calculated from 3-hourly winds from
12 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
the NZLAM 12 km regional atmospheric model5. The standard formula relating wind speed to surface stress involves a wind-speed-dependent term called the drag coefficient. For the present work this was calculated by the method of Smith (1988), however a comparison of preliminary model results with measurements indicated that wind-driven variability in the model was generally too low. The drag coefficient was therefore multiplied by a factor that was adjusted to optimise agreement: the final value chosen was 1.4. A similar adjustment has been found to be necessary in previous coastal modelling exercises around New Zealand by us (Hadfield and Zeldis 2012) and others (e.g. P. McComb pers. comm.). The need for this adjustment may indicate that the NZLAM wind speeds are biased low and/or that the Smith (1988) drag coefficient formula gives results that are too low for the wind and wave conditions in coastal areas.
The purpose of the Cook Strait model is to support a reasonably accurate description of the bathymetry of Cook Strait and, with it, the paths of currents through the strait.
To illustrate this point Figure 1-4 shows depth-average velocity vectors averaged over the last 730 days of a Cook Strait model run and Figure 1-5 shows similar data from the nested South Taranaki Bight model. A continuous current can be seen entering Cook Strait from the south along the Kahurangi coast, then crossing the strait at its shallower, western end. The path of the current then follows the 50–100 m depth band to the South Taranaki coast, skirting the Patea Shoals. From there it follows the coast south past Manawatu, Horowhenua and Kapiti, through the Narrows and then northward along the Wairarapa coast. The existence of this current system has been known for several decades, but the details of its spatial pattern and temporal variability were not previously well described. An accurate representation of this current and its variability is important because it is expected to have a major influence on sediment plumes from the proposed ironsand extraction operation.
5 NZLAM is part of the NIWA Ecoconnect environmental forecasting system: http://EcoConnect.niwa.co.nz/
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 13
Figure 1-4: Time-averaged, depth-average velocity from the Cook Strait model. Velocity vectors are averaged over 730 days and shown at every 4th grid point. Depth contours are at 50, 100, 250, 500, 1000 and 2000 m.
Figure 1-5: Time-averaged, depth-average velocity from the South Taranaki Bight model. Velocity vectors are averaged over 730 days and shown at every 4th grid point. Depth contours are at 10, 25, 50, 75 and 100 m.
1.4 South Taranaki Bight model The South Taranaki Bight model was used to generate the majority of the results in this report (Sections 5.1–5.4). It was forced at the lateral boundaries by data from the Cook Strait model at an interval of 3 hours. The South Taranaki Bight model had similar surface forcing and dynamics to the Cook Strait model, but with the addition of several processes:
14 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
. The ROMS sediment module was activated, representing both natural sediments and plumes generated by the project. More detailed information about the sediment module and associated parameterisations is given in Section 1.6 and the sediment properties are described in Section 3.
. Tidal forcing was applied at the boundaries. Amplitude and phase data for 13 tidal constituents were interpolated from the output of the NIWA EEZ tidal model (Walters et al. 2001). The ROMS tidal forcing scheme then calculated tidal sea surface height and depth-averaged velocity at each time step and added them to the lateral boundary data from the Cook Strait model. Applying tides in addition to the lateral boundary data from an outer model in this way means that the outer model fields do not need to be stored at a high temporal resolution.
. A bottom boundary layer scheme was activated (the Sherwood/Signell/Warner variant, see Warner et al. 2008). This scheme takes account of waves in calculating bottom drag and sediment erosion/deposition and requires data on the wave orbital velocity near the bottom.
. The larger rivers that drain into South Taranaki Bight were represented as point sources of freshwater and suspended sediment, as described in Section 1.5.
1.5 Patea Shoals model The Patea Shoals inner model was used in simulations where the emphasis was on short- range dispersion (Sections 5.5 and 5.6). In its setup and forcing it was very similar to the South Taranaki Bight model, except that there was no river input.
1.6 Sediment model setup The ROMS parameterisations for sediment and associated quantities (eg. bottom boundary layer, wave forcing) that were implemented on the South Taranaki Bight and Patea Shoals domains are as described by Warner (2008). The model accepts a series of user-defined sediment classes, each characterised by several properties such as grain size, grain density and settling velocity. Typically a continuous distribution of sediment particle sizes in reality is represented in the model by several sediment classes, each with a different grain size. A description of the sediment classes and their properties as used for the present report is given in Section 3.
At the bottom of the water column there is a multi-layer sediment bed, each layer being composed of a mixture of the sediment classes. The top bed layer exchanges sediment with vertically with the lowest level in the water column and horizontally with its immediate neighbours (a process called bedload transport).
The thickness of the bed layers is adjusted at each model time step according to the scheme described by Warner (2008), within any changes resulting in mass-conservative exchange between the layers. At the beginning of each time step an active layer thickness is calculated (Warner et al. 2008, Equation 21; Harris and Wiberg 1997, Equation 1). The active layer thickness sets a minimum for the top bed layer thickness, meaning that sediment is immediately mixed over this depth, and this is important in regulating the availability of fine sediment in the bed for erosion. It is important to note that the Harris and Wiberg relationship
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 15
for active layer thickness was developed for a location dominated by biogenic roughness rather than transport-induced bedforms.
For the simulations described below the total sediment bed thickness was set initially to 1 m, with eight layers. Initial layer thickness was 0.125 m, but this adjusts after the first time step. Incidentally, the ROMS model optionally allows for the depth at the base of the water column to be adjusted as the total sediment bed thickness changes, but this facility was turned off for the present simulations.
Vertical exchange of sediment between the top bed layer and the water column is the sum of two terms (Warner 2008, Equation 22). The first is deposition: it occurs continuously and is calculated separately for each sediment class as the product of the near-bottom concentration and the settling velocity. The second is erosion (Warner 2008, Equation 23; Ariathurai and Arulanandan 1978): it occurs only when the bottom stress exceeds a critical value, user-specified for each class.
Bedload transport (Warner et al. 2008 Section 3.4) is optionally calculated with the Soulsby and Damgaard (2005) formulation. This process has been included in only one of the simulations described below (the patch source simulation, Section 5.5) where it results in a modest increase, around 20%, in the rate at which medium sands are transported out of the patch area, relative to a simulation with bedload transport excluded.
Bottom stress is calculated with the Sherwood, Signell and Warner bottom boundary layer formulation (Warner et al. 2008, Section 3.7). This requires data on the height, period and direction of surface wind waves. Two sources were considered: the NZWAVE wave forecasting model6 and dedicated runs of the SWAN wave model (R. Gorman, pers. comm). Another choice to be made in the model setup was the bottom roughness formulation. A bottom roughness length is calculated from the median grain density of the sediment bed (Warner et al. 2008, Equation 44). This changes as the sediment bed evolves but for the simulations below it was typically ~0.4 mm. Additional terms account for sediment bedload transport (Warner et al. 2008, Equation 45) and for bedform ripples (Warner et al. 2008, Equation 46). The roughness length associated with bedload transport was negligible in these simulations. The bedform roughness length varied in space and time but was generally largest at around 20–30 m depth, where it was 0.5–2.5 mm. All these terms can be enabled or disabled in the ROMS code. As part of the calibration process, several simulations were carried out with different combinations of forcings and processes, the calibration target in this case being measured SSC data (Section 4). The model configuration finally selected used NZWAVE wave data and neglected the bedform roughness length. The main problem with the latter was that, when it was included, the model was not able to reproduce the isolated spikes that are seen by the ABS in near-bottom SSC (Section 4.3).
Figure 1-6 presents several time series relevant to erosion and deposition processes. The location is in a water depth of 31 m depth on outer Patea Shoals and appears elsewhere in this report as instrument site 7 (Figure 1-2a) and mining site A (Table 3-5). It is characterised by strong tidal currents, moderately large wave amplitudes and complex bedforms (Section 4.3 this report; MacDonald et al. 2012, Figure 3-56). It is one of 24 locations in the model
6 NZWAVE is part of the NIWA Ecoconnect environmental forecasting system: http://EcoConnect.niwa.co.nz/
16 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
where hourly data have been saved to allow evaluation of short-timescale phenomena like tides.
a)
b)
c)
Figure 1-6: Bottom boundary layer & sediment bed time series at instrument site 7. a) Near- bottom current speed; b) bottom wave orbital speed; c) sediment bed active layer thickness.
Panel a shows the near-bottom current speed: values are generally less than 0.4 m/s, with occasional excursions as high as 0.75 m/s. Panel b shows the near-bottom wave orbital
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 17
velocity, calculated within ROMS from NZWAVE wane height and period data. The values are frequently higher than the current speed (note the different vertical scale) with several excursions above 1 m/s. Panel c shows the active layer thickness calculated by ROMS. It is generally less than 10 mm, but occasionally exceeds 40 mm, with all the peaks coinciding with wave events. The figure suggests that erosion of bottom sediments here (and in shallower water) tends to occur in occasional high-wave events and that in the model these mix the upper sediment bed at this location to a depth of at least 40 mm.
1.7 River inputs Data for some 40 rivers were extracted from the NIWA Water Resources Explorer (WRENZ) website7 (Hicks et al. 2011) and 11 were selected for inclusion in the model based on their mean flow rate and sediment input rate (Table 1-1). The rivers that were selected comprised the ten with the highest flow rate, plus the Tangahoe River, which ranks 13th for flow but 9th for sediment input.
The model was supplied with a time series of daily-average flow rate and sediment input for each river. The flow rate was estimated from gauging station data collected by NIWA, Taranaki Regional Council and Horizons Regional Council. The gauging station data was available for all rivers except the Tangahoe, for which the WRENZ mean flow was used throughout. Where the gauging station was well upstream from the coast, the data were scaled by the ratio between the catchment area above the gauging station and the total catchment area. For several rivers, data were not available after Dec 2012; these gaps, plus a few short gaps elsewhere in the record, were filled with the WRENZ mean flow.
Table 1-1: Rivers represented in the inner model, with mean freshwater and sediment input rates from WRENZ.
Name Flow rate Sediment rate (m3/s) (kg/s) Whanganui River 229.0 149.03 Manawatu River 129.5 118.46 Rangitikei River 76.4 35.04 Whangaehu River 47.2 21.82 Patea River 30.4 9.85 Waitotara River 23.3 15.08 Otaki River 30.1 5.46 Whenuakura River 9.9 8.75 Kaupokonui Stream 8.6 0.31 Turakina River 8.4 9.54 Tangahoe 4.2 1.39 Total 593 373
For the Whanganui River, a time series of SSC was available up to December 2012 from Horizons Regional Council. A relationship between SSC and flow was established from the data before his time (Figure 1-7) and used to fill the remainder of the time series. (In this
7 http://wrenz.niwa.co.nz
18 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
case no filling of the flow time series was required, as a complete flow dataset was available.)
Figure 1-7: Daily average SSC vs flow for Whanganui River. Linear regression fit indicated by a red line.
For all rivers other than the Whanganui, SSC data was unavailable. Instead an SSC time series was constructed on the assumption that SSC is directly proportional to flow (as is approximately the case for the Whanganui), with a constant of proportionality adjusted so that the mean sediment input matched the WRENZ value.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 19
2 Hydrodynamic model evaluation
2.1 Field measurements A programme of field measurements was carried out in South Taranaki Bight, involving instrument deployments at several sites for three periods between September 2011 and July 2012. The programme is described in detail in a dedicated report (MacDonald et al. 2012). The data that are most relevant for the sediment plume modelling are:
. Vertical velocity profiles from acoustic Doppler current profilers (ADCPs) at five sites (Figure 1-2b).
. Profiles of temperature and salinity from moored sensors at several sites.
. Measurements of suspended sediment from optical and acoustic backscatter sensors (OBS and ABS).
Below, in the present section, modelled velocities are compared with the ADCP measurements. In Section 4.3 modelled near-bottom suspended sediment (sand) concentrations are compared with ABS data to give an approximate check on the model’s representation of bed resuspension processes.
The times and locations at which ADCP data are available are listed in Table 2-1.
Table 2-1: ADCP data availability from the field measurements. For more information see MacDonald et al. (2012), specifically Section 3.1 and Tables 3-1 & 3-2
Deployment Site 5 Site 6 Site 7 Site 8 Site 10 D1, 06/09/2011 to 01/12/2011 X X X D2, 08/12/2011 to 09/02/2012 X X X D3, 24/04/2012 to 01/07/2012 X X
In all the deployments in South Taranaki Bight, the ADCP instruments were mounted on the bottom, pointing vertically upwards and measuring horizontal velocities every 10 minutes at a series of levels (or “bins”) above the instrument. The vertical spacing between the bins varied between instruments but was typically 0.5 m. The model was set up to store velocity profile data at all the ADCP locations at an interval of 30 minutes. For comparisons between ADCP and model, the model profiles were interpolated vertically and in time to match a specified ADCP level; the analyses below concentrate on one that will be labelled “mid-depth”, i.e. the one nearest to halfway between the surface and the bottom. Also, we consider the tidal and sub-tidal components separately, the former being estimated by fitting harmonics of specified frequency to the data and the latter by applying a low-pass temporal filter to the data, an operation known as detiding.
2.2 Tidal current comparison This section considers the accuracy of the model’s representation of tidal currents. As is the case elsewhere around New Zealand, the dominant tidal constituent in the area is the lunar, semi-diurnal constituent (M2), with a period of 12.42 hours. Figure 2-1 compares measured
20 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
and modelled M2 tidal ellipses for one of the ADCP datasets, namely Site 7 (outer Patea Shoals), Deployment 2.
Figure 2-1: M2 tidal velocity comparison (ADCP Site 7 Deployment 2). Mid-depth M2 tidal ellipses from ADCP (blue) and model (red). The axes correspond to the velocity components towards due east (u) and due north (v). The ellipses represent the magnitude and orientation of the tidal velocity variations (see text) and the straight line from the origin to the ellipse represents the phase.
The tidal ellipse shows tidal velocity variations: it represents the path taken by the tip of a tidal current vector, rotating at a constant angular frequency and changing in length (current speed) through a tidal cycle. The tidal ellipse is defined by four parameters:
. Semi-major amplitude (m/s): The semi-major axes are lines from the origin to the two most distant points on the ellipse perimeter. The two axes are equal in length, and this length represents the amplitude of the velocity along the semi- major direction, or the maximum current speed during a tidal cycle.
. Eccentricity: At right angles to the semi-major axes are the semi-minor axes, representing the minimum current speed during a tidal cycle. The eccentricity, or “fatness”, of the ellipse is the ratio of semi-minor to semi-major axis lengths. The eccentricity can be positive (vector rotates anti-clockwise) or negative (clockwise).
. Inclination (°T): The inclination is the orientation of one of the semi-major axes. The choice between the two is arbitrary: here we take the semi-major axis directed towards the northeastern or southeastern quadrant and express the inclination as the orientation in degrees clockwise from true north.
. Phase (°): The phase relates to the time at which the rotating tidal current vector passes through the semi-major axis. A phase difference of 1° corresponds to a time difference of 1/360th of the tidal period.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 21
The tidal ellipses shown in Figure 2-1 clearly match reasonably well in magnitude and orientation. Similar ellipses for all ADCP deployments are shown in Figure C-1 and the ellipse parameters are tabulated in Table C-1. Overall agreement is very good. The semi- major amplitudes agree to within ±6%; the eccentricities agree within ±0.04, the inclinations within ±8°, and the phases within ±2.4° (5.0 minutes). This is a very good match.
The tidal current amplitude normally decreases towards the bottom due to friction. Figure 2-2 compares measured and modelled vertical profiles of the M2 semi-major amplitude for the same ADCP dataset as Figure 2-1. The agreement in the middle and upper water column is very good (as was evident from Figure 2-1). Measured and modelled amplitudes both decrease towards the bottom, as expected. The lowest level at which data are available from the ADCP instrument at this site is 2.05 m above the bottom: here the modelled amplitude exceeds the measured amplitude by 10%. Similar profiles are shown for all ADCP deployments in Figure C-2. Agreement is generally good, with a tendency for modelled amplitude to exceed measured amplitude by a modest amount in the lowest few metres. This may indicate that the effective bottom roughness is somewhat lower in the model than in reality.
Figure 2-2: M2 tidal current profile comparison (ADCP Site 7 Deployment 2). Semi-major amplitude of the M2 tide versus height above the bottom from ADCP (blue) and model (red).
The M2 constituent dominates the tidal velocity but several smaller-amplitude constituents also contribute. The larger ones in this area are S2 (12 hours), N2 (12.66 hours), and K1 (23.93 hours), for which mid-depth tidal ellipses are compared in Figure 2-3. Note the different velocity scale in this figure relative to Figure 2-1. Agreement is good, though not as close as for M2. The largest mismatch evident in these graphs is a difference of 31°, or 2.1 hours, in the phase of the K1 constituent. This difference probably results from a bias in the
22 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
lateral boundary data taken from the EEZ tidal model, which is known to have significant errors for this constituent in this region (Stanton et al. 2001).
Figure 2-3: Tidal velocity comparison for S2, N2 and K1 (ADCP Site 7 Deployment 2). Mid-depth tidal ellipses from ADCP (blue) and model (red).
2.3 Sub-tidal current comparison A comparison of sub-tidal currents is shown in Figure 2-4. The upper panel in this figure shows a scatter plot of velocities, with the ellipse in this case being a variance ellipse, a conventional representation of the magnitudes of variability in velocity data. A variance ellipse can be characterised by its semi-major axis (in this context called a principal axis), eccentricity and inclination, like a tidal ellipse. However a variance ellipse does not have a phase (since it says nothing about the timing of the variability) and its eccentricity has no sign (since it says nothing about the rotation of velocity vectors). Also, the centres of the variance ellipses in Figure 2-4 are offset from the origin by an amount representing the mean current
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 23
over the period of the deployment. The lower two panels in the figure indicate how well the fluctuations in sub-tidal velocity match up in time. The centre panel shows fluctuations along the principal axis of maximum variability (this being a compromise between the principal axes of the observed and modelled variance ellipses) and the bottom panel shows fluctuations perpendicular to this axis.
Figure 2-4 indicates good agreement between model and measurements for the ADCP deployment in question. The variance ellipses have a similar shape and orientation, there is a high correlation (r = 0.849) between measured and modelled fluctuations along the principal axis and a somewhat lower correlation (r = 0.555) perpendicular to this. The mean current (the offset of the centre of the ellipse) is small relative to the dimensions of the ellipse but appears to be directed to the east in both cases.
On a technical matter, the time series in Figure 2-4 are quite smooth, as a result of the application of a low-pass filter in detiding the data. (The filter is the 51G113 filter from Thompson, 1983, applied to hourly values; see Figure 2 of that article for the filter’s frequency response.) To investigate the effect of this smoothing, the model-measurement comparison has been repeated with an alternative method, namely, removal of the tides by analysing a set of several tidal constituents (3 semi-diurnal and 2 diurnal) from the data and then constructing and subtracting the tidal time series. A comparison of the results of the two methods (Figure 2-5, along with other output not shown) indicates that the latter method of detiding does leave more high-frequency information in the data, but for the purposes of comparison between model and measurements, both methods give similar results, as long as the model and measurement time series are detided in the same way.
24 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Figure 2-4: Sub-tidal velocity comparison (ADCP Site 7 Deployment 2). All graphs are based on time series of mid-depth detided velocity: a) Scatter plots and variance ellipses from ADCP (left-hand pane) and model (right-hand pane); b) Velocity components from ADCP (blue) and model (red) along the major axis of the variance ellipse; c) as (b) but for velocity components perpendicular to the major axis.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 25
Figure 2-5: Comparison of detiding methods (ADCP Site 7 Deployment 2). The along-axis time series comparison of Figure 2-4 detided with the 51G113 low-pass filter (upper panel) and tidal analysis and removal (lower panel).
A comprehensive set of comparisons for the mid-depth sub-tidal currents is presented in Appendix C. Variance ellipses are compared in Figure C-3, along-axis time series in Figure C-4 and across-axis time series in Figure C-5; the numeric values are tabulated in Table C-2. In assessing the degree of agreement between the measured and modelled mean currents, one should remember that in several cases the mean currents are small relative to the variability. Both measurements and model indicate a mean flow directed towards east- southeast (90–110°) at all locations except Site 7, where the mean current is directed to the northeast (35–75°). Modelled and measured directions agree to within ±10°, with a couple of exceptions where the magnitude of the mean measured current is very small. Modelled and measured magnitudes agree to within ±0.03 m/s, with the exception of Site 8 Deployment 3, where the modelled magnitude (0.09 m/s) is 1.5 times the measured magnitude (0.06 m/s). Comparing measured and modelled values of semi-major axis length, the model under- predicts this quantity by ~35% at the Site 5, ~30% at Site 6, and 0–12% at the other sites. The eccentricity and inclination of the variance ellipse are very close for all deployments (eccentricity within ±0.09, inclination within ±7°). When a comparison is made between measured and modelled velocity components, along and across the principal axis, the correlation is high for the along-axis components in all cases (r = 0.85–0.93) and lower for
26 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
the across-axis component. For Site 5 the cross-axis correlations are small, however the low eccentricity of the variance ellipses at this site implies that the across-axis currents are very weak; at the other sites the cross-axis correlations are larger (r = 0.55–0.84).
The same set of comparisons has been done for sub-tidal currents 5 m below the surface and at the lowest ADCP level. Near the surface (Table C-3), the overall pattern of the comparison is similar to the mid-depth comparison. The currents (in terms of mean magnitude and semi-major amplitude) are generally somewhat stronger near the surface and temporal correlations slightly higher. The ratio between modelled and measured semi-major amplitude is higher, and is now around 1.1 at the outer sites (7–10). Near the bottom (Table C-4) the same differences are seen in reverse. Compared to the mid-depth, currents are somewhat weaker, temporal correlations slightly lower, and the model tends to under-predict the semi-major amplitude a little more. For two datasets (Site 6, Deployment 2 and Site 7, Deployment 2) the modelled and measured mean directions differ greatly by 169° and 84°, respectively, but in both cases the mean current (0.01 m/s) is small relative to the variability (semi-major amplitude 0.05–0.08 m/s).
The degree of agreement here between the modelled and the measured velocities is as good as, or better than, that achieved in other modelling exercises on the New Zealand continental shelf (e.g. Hadfield and Zeldis 2012). The major reservation relates to the wind-driven variability: a relatively large factor (1.4) has been applied to the wind stresses, and with this factor the model underestimates the variability (semi-major axis amplitude) at the sites closest to the coast (Sites 5 and 6). The most likely explanation is that the surface wind data from the NZLAM atmospheric model do not fully capture the channelling and acceleration of winds through Greater Cook Strait. However confirmation of this explanation would require data from a higher-resolution atmospheric model. The implications of the model underestimating the wind-driven variability near the coast on sediment transport are not entirely clear, but it is likely that the model will underestimate the extent to which river plumes will be spread along the coast.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 27
3 Sediment properties and release parameters The ROMS sediment scheme requires the modeller to define a set of sediment classes and, for each one, to specify several properties, notably: median grain size; grain density; porosity (when in the sediment bed) ; settling velocity (when in suspension); critical bed shear stress for erosion and deposition; and a rate parameter used in the formula relating erosion rate to bed shear stress. The number of sediment classes is unlimited, but the computational expense increases for each additional class.
The results presented in this report were produced with the ROMS sediment model in on one of the inner domains. The following sub-sections describe the different model configurations
3.1 Natural sediments The base simulation represented natural sediment processes using 7 sediment classes:
. The river-derived sediments are injected by the rivers. There are two classes: a coarse silt (16–63 µm) and a fine silt/clay (< 16 µm).
. The seabed-derived sediments comprise the seabed at the beginning of the simulation. They range from a coarse sand (500–1000 µm) to a fine silt (16– 63 µm).
There were two reasons for introducing the natural sediments:
. To support a realistic interaction between the mining-derived sediment plumes and the seabed;
. To produce estimates of natural suspended sediment concentrations that are at least approximately realistic and can be compared with estimates of sediment concentrations resulting from the ironsand extraction operation.
The properties of the natural sediment classes are given in Table 3-1. Other properties required by ROMS include an erosion rate parameter (2 × 10−4 kg m−2 s−1 for all classes) and a porosity (0.4 for all classes).
Table 3-1: Natural sediment classes and properties.
Source Nominal Grain size Grain Settling Critical size range (µm) density velocity stress (µm) (kg m−3) (mm/s) (Pa) River 16–63 31 2650 0.63 0.100 River 4–16 8 2650 0.10 0.100 Seabed 500–1000 710 3100 103 0.431 Seabed 128–500 256 3100 38 0.219 Seabed 63–128 90 3100 6.3 0.160 Seabed 16–63 31 3100 0.76 0.102 Seabed 4–16 8 3100 0.10 0.100
28 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
The properties of each sediment class were based on data and formulas in the literature. For each class a characteristic settling velocity and threshold bed shear stress was calculated from the mid-point grain size according to formulas given by van Rijn (1984) and Soulsby (1997). The settling velocity of seabed and river silts was limited to a minimum of 0.1 mm/s, to allow for the fact that very fine sediments suspended in the sea are normally in flocculated form. The critical stress was limited to a minimum of 0.1 Pa, which is characteristic of unconsolidated mud beds (Ahmad et al. 2011).
The sediment input rate of each river varies with time and is estimated by the methods described in Section 1.5; this is partitioned as 50% coarse silt and 50% fine silt/clay. This assumption is based on a few size-fractionated SSC measurements from the Whanganui River in flood. It is a reasonable assumption for flood conditions, but in low-flow conditions it probably overestimates the coarse silt fraction. However the great majority of the riverine sediment input occurs in flood conditions, so the assumption should be acceptable.
The seabed is initially populated with a combination of coarse sand (500–1000 µm, 20%), fine–medium sand (128–500 µm, 72%), very fine sand (63–128 µm, 6%), coarse silt (16– 63 µm, 1.5%) and fine silt (4–16 µm, 0.5%). The proportions were initially based on seabed particle size distribution (PSD) data from the extraction area, and the fine sediment fractions were then adjusted so that the model produces surface SSCs of approximately the correct magnitude in the near-shore area (Sections 4.1 and 4.2).
As described in Section 1.3, the model of natural sediment processes was initialised at 2000 days (relative to a reference time of 00:00 UTC on 1 January 2005) and run to 3000 days, with statistical analyses based on the last 730 days of the simulation. The sediment bed adjusts within the first 100–200 days after initialisation as the finer classes are stripped out of the uppermost bed layer in high-energy areas like Patea Shoals but remain in low-energy areas. A slower adjustment process occurs throughout the duration of the simulation as seabed material is slowly eroded from the high-energy areas and deposited in the low-energy areas.
3.2 Mining-derived sediments Two main sediment streams from the ironsand extraction are considered: the hydro-cyclone overflow discharge and the de-ored sand discharge. A third source, not of sediment but of fresh water from hyperbaric filters used to dewater the iron ore concentrate is also described below. The hydro-cyclone overflow (Table 3-2) results from dewatering of the de-ored sand before it is pumped to the bottom. It is a discharge of mostly fine sediment with a large flow (8.8 m3/s) of water, released approximately 10 m above the bottom. (The exact height and discharge parameters are not known.) The de-ored sand discharge (Table 3-3) involves de- watered, de-ored sand being released from a pipe 4 m above the bottom, with a view to depositing it as compactly as possible, usually into a pit that has been excavated earlier. The de-ored sand is predominantly fine–medium sand (125–500 µm) with some finer material.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 29
Table 3-2: Hydro-cyclone overflow discharge rates.
Size range (µm) Mass flux (tonne/h) Mass flux (kg/s) < 8 102.6 28.5 8–16 73.0 20.3 16–38 92.5 25.7 38–90 88.7 24.6 90–125 28.2 7.8 125–150 7.6 2.1 150–212 6.7 1.9 > 212 0.0 0.0 Total 399.3 110.9 Total (< 90 µm) 356.8 99.1 Total (< 38 µm) 268.1 74.5
These sediment streams are represented in the model by two different mechanisms, treated in different simulations. The first is a continuous source of fine suspended sediment at two different heights, one 15 m below the surface (representing the hydro-cyclone discharge) and the other 1.5 m above the bottom (representing the finer material in the de-ored sand discharge). The second is a 3 × 2 km rectangular patch of sand representing the area filled by one year’s mining and containing all the material that has not been released in the suspended source. The set-up of the simulations that implement these two sources is described further below.
The partitioning of the hydro-cyclone overflow and de-ored sand streams between the suspended source and patch source requires some consideration. It is assumed that all the material in both streams with a grain size > 90 µm will be deposited on the bottom shortly after release, where it will be buried in the de-ored sand patch. Meanwhile, it is assumed that all material in both streams with a grain size < 38 µm will be released into suspension. Some of this finer material may be deposited after release, but typically not close enough to the discharge point to be buried.
30 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Table 3-3: De-ored sand discharge rates.
Size range (µm) Mass flux (tonne/h) Mass flux (kg/s) < 8 11.8 3.3 8–16 8.5 2.4 16–38 11.5 3.2 38–90 181.2 50.3 90–125 250.7 69.6 125–150 337.0 93.6 150–212 1006.7 279.6 212–250 1067.1 296.4 250–355 2073.5 576.0 355–500 925.5 257.1 500–710 329.7 91.6 710–2000 314.3 87.3 > 2000 320.0 88.9 Total 6837.4 1899.3 Total (< 90 µm) 213.0 59.2 Total (< 38 µm) 31.8 8.8
The partitioning of the material with a grain size that falls between these two limits, the 38– 90 µm class, was addressed with the aid of simulations with a high-resolution computational fluid dynamics (CFD) model of the de-ored sand plume carried out by Svasek Hydraulics (MTI Holland 2013a,b). With this model, for a range of scenarios representing high-energy conditions, it was estimated that of the 50.3 kg/s of 38–90 µm material discharged in the de- ored sand (see ) between 32 and 43 kg/s would still be in suspension 100 m downstream from the source, with the remainder deposited on the bottom. For the present modelling, it was assumed that the 38–90 µm class in the de-ored sand had a release rate of 35.0 kg/s in the suspended source, with the remainder being deposited at a rate of 15.3 kg/s.
3.2.1 Suspended source Based on the above information and calculations, 8 sediment classes have been defined to represent the suspended sediment source from the ironsand extraction. Four are associated with the hydro-cyclone overflow and another four with the de-ored sand discharge. The de- ored sand classes are identical to the respective hydro-cyclone classes, but are tracked separately in the model to allow material from the two sources to be distinguished. As with the natural sediment classes, properties like fall speed and critical erosion stress have been calculated from formulas in the literature. Unlike the natural sediment classes, a lower limit of 0.1 mm/s is not imposed on the fall speed, because the mining-derived sediment is not expected to flocculate as readily as natural sediment. The suspended-source sediment classes, their properties and their discharge rates are listed in Table 3-4.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 31
Table 3-4: Sediment classes, properties and discharge rates for the ironsand extraction suspended source.
Source Nominal Grain size Grain Settling Critical Source size range (µm) density velocity stress rate (µm) (kg m−3) (mm/s) (Pa) (kg s−1) Hydro-cyclone 38–90 58 3100 2.8 0.133 24.6 Hydro-cyclone 16–38 25 3100 0.52 0.100 25.7 Hydro-cyclone 8–16 11 3100 0.10 0.100 20.3 Hydro-cyclone < 8 4 3100 0.01 0.100 28.5 De-ored sand 38–90 58 3100 2.8 0.133 35.0 De-ored sand 16–38 25 3100 0.52 0.100 3.2 De-ored sand 8–16 11 3100 0.10 0.100 2.4 De-ored sand < 8 4 3100 0.01 0.100 3.3
The suspended source was simulated in a model that also included the 7 natural sediment classes (Section 3.1), giving a total of 15 sediment classes in all. The suspended source was turned on at 2200 days. seventy days before the beginning of the two-year period over which statistics were calculated. Based on advice from the client, the source was operated on a 20- day cycle, first on for 16 days at the rates given in Table 3-2, then off for 4 days, giving an 80% up-time. (The up-time assumed for operational planning is approximately 70%, so the 80% assumed here is conservative for the purpose of plume modelling.) The source was also turned off when the significant wave height at the site exceeded 4 m, which occurs approximately 2.5% of the time, based on the wave data used to drive the model.
The total discharge rate of sediment < 90 µm in the suspended source is 143 kg/s. Allowing for 80% uptime, this implies an annual mean of 114 kg/s. For comparison, the annual mean sediment discharge for the Whanganui River estimated by WRENZ is 149 kg/s and the total for all the rivers in Table 1-1 is 373 kg/s.
Two source locations are considered, as specified in Table 3-5.
Table 3-5: Source locations.
Label Location Water depth (m) Description Longitude Latitude A 174.194°E 39.850°S 30.8 Inner end of mining area, at measurement site 7 B 174.070°E 39.888°S 42.0 Outer end of mining area, at measurement site 10
3.2.2 Suspended source recovery simulation A further simulation was carried out with the suspended source to investigate how long its effects would persist after the release of sediment finishes. This could not be done by turning the source off at (say) 3000 days and continuing for another year or two, because there are no forcing data available after approximately 3150 days (the date of writing of this report). To work around this problem the sediment fields (in suspension and in the bed) were translated
32 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
back in time to 2200 days used in a model restart, combining the time-shifted sediment fields with the hydrodynamic fields at this time. (This procedure is not entirely self-consistent as it results in a sediment field that is not exactly in balance with the hydrodynamic fields, but it does avoid “hydrodynamic shock’ at the restart.) This simulation was carried out for source location A only.
3.2.3 Patch source To model the fate of de-ored sand buried at the mining site we consider a rectangular patch representing one year’s worth of ironsand extraction and populate this patch with material that reflects the composition of the combined hydro-cyclone and de-ored sand discharge streams, minus all the material that was released in the suspended source (Section 3.2.1).
The area of this source is calculated as follows: Assuming a volume extraction rate of 1.195 m3/s at full operation (mass extraction rate 8000 tonne/h with a bulk density of 1860 kg/m3) and an up-time of 80%, the annual volume extracted is 30.15 × 106 m3. At a mean mined depth of 5 m, this implies that an area of 6.05 km2 would be mined in one year. (There are various sources of uncertainty in this number but the largest is the mean mined depth, which could differ from the assumed value by a factor of two.) This area is represented in the model as a 3 × 2 km rectangular patch centred on the mining site, which is taken to be mining site A.
The sediment classes for the patch source are listed in Table 3-6. There is no < 38 µm class because all this material is assumed to have been released in the suspended source. The fractions in the right-hand column sum to 100% and are in proportion to the deposition rates.
Table 3-6: Sediment classes and composition for the deposited patch source.
Source Size range Grain size Deposition rate Fraction (µm) (µm) (kg/s) Patch 500–1000 710 268 14.3% Patch 125–500 256 1507 80.7% Patch 90–125 107 78 4.2% Patch 38–90 58 15 0.8%
The material in the seabed around the patch is assumed to be unmined sand, with a composition based on data from the client describing the run-of-mine (ROM) feed. The sediment classes (Table 3-7) are the same as the patch classes, with the addition of a 5th < 38 µm class. The fractions in the right-hand column are in proportion to the mining rates. Compared to the surrounding seabed, the patch has a lower proportion in the 38–90 µm range, and none at all < 38 µm, with somewhat more in the remaining size ranges.
The simulation was run on the smaller (Patea Shoals) inner domain with no river input. (This is a reasonable approximation as the footprint of the patch source tends to be limited to a small area and there should be no interaction with the rivers.) The model was initialised at 2000 days with seabed sediments (Table 3-7) filling the domain, then at 2200 days the 3 × 2 km patch area was replaced with the patch-source sediments (Table 3-6). The simulation was continued to 3000 days.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 33
Table 3-7: Sediment classes and composition for the seabed surrounding the deposited patch source.
Label Size range Grain size Mining rate (kg/s) Fraction (µm) (µm) Seabed 500–1000 710 268 14.2% Seabed 125–500 256 1507 79.7% Seabed 90–125 107 62.8 3.3% Seabed 38–90 58 22.5 1.2% Seabed < 38 9 30.1 1.6%
Unlike the other simulations described in this report, the patch simulation included bedload transport. Comparison with an earlier simulation without bedload transport indicates that its effect is to produce a modest increase in the movement of patch sands onto the surrounding seabed.
The original motivation in setting up the patch source was to estimate the change in suspended sediment concentrations resulting from the difference in sediment properties between the patch and the surrounding seabed. (As noted above, the main difference is that fine sediments are lower in the patch.) However it is likely for several reasons that the deposited patch would be eroded more readily than the surrounding seabed:
. The patch material when it is first deposited will be unconsolidated, with a higher porosity than the surrounding seabed.
. Any biostabilisation in the surrounding seabed will not be active in the patch.
. The patch surface may sit at a different level from the surrounding seabed, leading to turbulence generation and potentially faster erosion.
. The patch will be depleted of the very high density titanomagnetite grains, which may act to stabilise the seabed against erosion. (It has also been suggested that titanomagnetite may stabilise sediment beds by means of its magnetism, but this suggestion is not well supported—K. Bryan pers. comm.).
In view of these reasons, analyses of the patch simulations have considered only the patch material itself, to give an order-of-magnitude estimate of how readily this is eroded, transported and deposited.
3.2.4 Freshwater source The freshwater source consists of freshwater that has been used to rinse the iron ore concentrate before shipment and subsequently removed by hyperbaric filters. The flow rate is assumed to be 0.198 m3/s, and the source is marked in the model with a passive tracer allowing this water to be tracked. The concentration of any dissolved salts is not known, but when this information becomes available the tracer concentration can be scaled as necessary (providing the dissolved salts are not present in high enough concentrations to affect the density significantly). The freshwater is released approximately 2 m below the surface.
34 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
4 Sediment model evaluation
4.1 Surface SSC comparison with remote-sensed data Figure 4-1 shows the mean surface concentration of natural sediments from the model in comparison with an estimate from remote sensing (Pinkerton et al. 2013). In both cases there is a band of elevated SSC adjacent to the coast. The width of this band is approximately 20 km and it is widest over the relatively shallow water of the Patea Shoals. Outside the coastal band the modelled mean SSC (~ 0.3 mg/L) is somewhat lower than the remote sensed estimate (~ 0.7 mg/L). Adjacent to the coast the modelled mean (~ 20 mg/L) is somewhat higher than the remote-sensed estimate (~ 10 mg/L).
Note that South Taranaki Bight model receives no input of sediment from outside the domain In fact we know from satellite images that visible sediment plumes often extend from the southern side of Cook Strait, apparently associated with sediment input from the South Island coast and rivers. The maximum (green area) in the southwestern corner of the remote-sensed mean supports that interpretation. Therefore it is likely that the low modelled surface SSCs in the deeper water are underestimates due to the neglect of this input.
The relationship by which SSC has been estimated from remote sensing data has considerable scatter (Pinkerton et al. 2013, Section 4-1, Figure 4-1c) so the accuracy of the remote-sensed mean is uncertain, particularly at the high and low end. Furthermore, satellite remote sensing only returns data in cloud-free conditions, and the conditions in which surface SSC is highest (large waves, high river flow) tend to occur (though not exclusively) in cloudy conditions. With these limitations in mind, the degree of agreement between model and remote sensing is very satisfactory.
Figure 4-1: Mean surface concentration of natural sediment. Model (left) and estimate from 10 years of satellite data (right, redrawn from data shown in Pinkerton et al. 2013, Figure 4-1).
4.2 Surface SSC comparison with in situ measurements A programme of field measurements of near-shore optical water quality was carried out in February to May 2013 (MacDonald et al. 2013). It included continuous measurements of several optical parameters, including turbidity, for 6 weeks at 6 sites along the coast between Whanganui and Ohawe. The sites were all at approximately 10 m water depth and the optical sensors were 3 m below the surface. A relationship between turbidity and SSC was
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 35
established using data from two synoptic boat surveys and with this relationship the continuous turbidity measurements were converted to SSC (MacDonald et al. 2013 Figures 3-17a, 3-19a, 3-21a, 3-23a, 3-25a and 3-27a).
Comparisons between the observed SSC data and modelled SSC at the same locations and times (Figure 4-2 to Figure 4-7) show a generally good correspondence between the two datasets in terms of the timing of the peaks in SSC, with a tendency evident for the model to overestimate SSC by a factor of up to two. For a highly variable quantity like SSC this represents good agreement.
As noted by MacDonald et al.(2013) the near-shore optical measurements were carried out during a period of low river flow, with generally low near-shore SSC. Shortly after the end of the model runs (3000 days, 20 March 2013) there was an increase in river flows with an increase in observed SSC. It would be desirable to extend the model runs to mid-June 2013 to see if the model reproduces this increase. This would require several model forcing datasets to be updated and has not been undertaken at the time of writing of this report.
36 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Figure 4-2: SSC time series at near-shore site 11 (Whanganui). Modelled SSC (red) versus measured time series (blue, cf. MacDonald et al. Figure 3-17a).
Figure 4-3: SSC time series at near-shore site 12 (Kai Iwi). Modelled SSC (red) versus measured time series (blue, cf. MacDonald et al. Figure 3-19a).
Figure 4-4: SSC time series at near-shore site 13 (Waitotara River). Modelled SSC (red) versus measured time series (blue, cf. MacDonald et al. Figure 3-21a).
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 37
Figure 4-5: SSC time series at near-shore site 14 (Patea). Modelled SSC (red) versus measured time series (blue, cf. MacDonald et al. Figure 3-23a).
Figure 4-6: SSC time series at near-shore site 15 (Manawapou). Modelled SSC (red) versus measured time series (blue, cf. MacDonald et al. Figure 3-25a).
Figure 4-7: SSC time series at near-shore site 16 (Ohawe). Modelled SSC (red) versus measured time series (blue, cf. MacDonald et al. Figure 3-27a).
38 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
4.3 Near-bottom SSC comparison with in situ ABS data This section presents a comparison of modelled near-bottom sand (> 63 µm) concentrations with ABS data from instrument sites 6, 7, 8 and 10 on Patea Shoals (Figure 1-2b). In the Oceanographic Measurements report the ABS data are presented in terms of a quantity called Cref, the concentration at a reference level (MacDonald et al. 2012, Figures 3-45, 3-55, 3-57 and 3-58). For comparison with the model they were converted to average concentrations over the lowest 1 m of the water column (Iain MacDonald, pers. comm.), which more closely matches what the model calculates.
The model produces a series of discrete resuspension events, which matches what the ABS observes. However the model does not reproduce the large difference in observed SSC between site 7 (where observed SSC reaches 500 mg/L on occasion and is generally under- predicted by the model), site 6 (where observed SSC exceeds 100 mg/L on only one occasion and is generally over-predicted by the model) and the deeper sites, 8 and 10, where observed SSC is less than 20 mg/L and is over-predicted by the model). The modelled near-bottom sand concentration generally peaks at around 100 mg/L at the shallow sites and 60–80 mg/L at the deeper sites.
The large difference in sand concentration measured by the ABS between the shallower sites, 6 and 7, was noted by MacDonald et al. (2012) and attributed to the complex bedforms around Site 7 (their Figure 3-56). They quote Green and Black (1999) who show that, given the same wave forcing, differences in bed form geometry can give rise to concentration differences of a factor of 20. Meanwhile at the deeper sites, biostabilisation by polychaete tube worms may be an important factor stabilising the bed (Beaumont et al. 2013).
Our conclusion from the comparison is that the sediment model does not reproduce the wide range of variation in susceptibility to sand resuspension between different locations on Patea Shoals, and furthermore could not do so without a lot of tuning to local conditions. The formulation of resuspension is likely to overestimate resuspension in some places and underestimate it in others.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 39
Figure 4-8: Time series of near-bottom sand concentration at instrument site 6. Modelled sand concentration (red) versus estimate from ABS measurements.
Figure 4-9: Time series of near-bottom sand concentration at instrument site 7. Modelled sand concentration (red) versus estimate from ABS measurements.
Figure 4-10:Time series of near-bottom sand concentration at instrument site 8. Modelled sand concentration (red) versus estimate from ABS measurements.
40 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Figure 4-11:Time series of near-bottom sand concentration at instrument site 10. Modelled sand concentration (red) versus estimate from ABS measurements.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 41
5 Sediment plume results
5.1 Statistical measures of SSC The following subsections present maps of statistical parameters calculated from the suspended sediment concentrations (SSC), considering both the natural sources (river and seabed) and the additional sediments introduced by the ironsand extraction operation. The effect of the additional mining-derived sediments is assessed by comparing simulations with only natural sediments against simulations with both natural and mining-derived sediments. The different sediment types are tracked separately, but ultimately what is of interest is the difference in total sediment concentration (or, for a later report, sediment optical effects) between the simulations with and without the additional mining-derived sediments.
It turns out that this assessment is more complicated than would appear at first glance because there are interactions in the model between the different sediment types. This is illustrated by Figure 5-1, which shows 4 model-generated surface SSC time series from site 11 (Whanganui) of the near-shore optical programme (MacDonald et al. 2013). This is the site for which modelled and observed SSC is compared in Figure 4-2. The conclusion from this comparison was that the model reproduces observed SSC reasonably well, perhaps over-estimating by as much as a factor of 2. The 4 time series are all shown for the duration of the model run (t = 2000 to 3000 days). The vertical, dashed red line at 2270 days represents the beginning of the two-year period over which statistical parameters were calculated. Those parameters are indicated by horizontal, dashed red lines: they are the median, the 95th and 99th percentile, and the maximum. The median, or 50th percentile, is the value that is greater than the time series data 50% of the time. It will be used later to represent typical SSCs. The 95th and 99th percentiles are the values that exceed the data 95% and 99% of the time or, conversely, the values that are exceeded 5% and 1% of the time, respectively. The 99th percentile will be used as a robust indicator of the high end of the distribution of SSCs: on average it will be exceeded on 3.65 days per year. Also shown is the maximum, the highest value in the time series.
Panel a of Figure 5-1 shows the natural SSC, from a simulation with natural sediment only. The time series is very “spiky”, i.e. for most of the time the SSC is relatively low, ~10 mg/L or less, but for brief periods it increases to much larger values, around 50–200 mg/L. This spikiness is typical of SSC data. It means the frequency distribution of SSCs is skewed, with a peak at low values and a long tail at high values. The median for this time series is 5.3 mg/L and the 99th percentile is 122.6 mg/L. The large ratio between these two numbers (23.1) is a manifestation of the skewness of the distribution.
The remaining panels of Figure 5-1 show time series from a simulation with both natural and mining-derived sediments. Panel b shows only the natural SSC from this simulation. The time series data in this panel follow the data in panel a quite closely (the peaks occur at the same times) but there are differences in the statistical parameters. From panel a to panel b the median is reduced from 5.3 to 4.6 mg/L, the 99th percentile is also reduced, but very slightly, from 122.6 to 122.4 mg/L, and the maximum is increased by a surprisingly large amount, from 188 to 218 mg/L. Why do these differences occur? There are two reasons:
. The sediments in the model affect the currents, though normally by a small amount, by altering the water density and also the bottom roughness (which
42 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
affects bottom drag). Any change, however small, in a model like this then affects the transport processes, which can have a large effect on the concentration at any given point in space and time. In this sense the model— like the real world—displays “chaotic” behaviour.
. The different sediments, once deposited on the bottom, then compete with each other during resuspension. It appears that overall (and with exceptions) the introduction of mining-derived sediment tends to reduce the concentration of suspended natural sediments.
Panel c in Figure 5-1 shows the concentration of mining-derived sediment alone from the same simulation as panel b. The release of the mining-derived material starts at 2200 days and it takes 15 days or so before it is first seen at this site and another 50 days or so to approach its highest values. At this site, the mining-derived SSC is generally much lower (median 0.6 mg/L; 99th percentile 3.6 mg/L) than the natural SSC. Panel d shows the total SSC. In principle the net effect of the additional mining-derived sediment on the SSC can be calculated by taking the differences in the respective statistical parameters between panels a and d. By these calculations: the median has stayed the same at 5.3 mg/L; the 95th percentile has increased from 57.2 to 59.8 mg/L; the 99th percentile has increased from 122.5 to 124.6 mg/L; the maximum has increased from 188.0 mg/L to 220.6 mg/L. The increase in the maximum is moderately large as a percentage, but appears to be a manifestation of the chaotic behaviour of the model. For all the other parameters the effect of the mining-derived sediment is small in relation to existing sediment concentrations.
Figure 5-2 is the same as Figure 5-1, but for a site 18 km from the coast and 8 km from mining source A, frequently in the path of the mining plume. Here the natural sediment SSCs (panel a) are lower (median 0.8 mg/L; 99th percentile 8.2 mg/L) and the mining-derived SSCs (panel c) are higher (median 2.7 mg/L; 13.4 mg/L). Here the net effect of the mining sediments in raising the total SSCs is clear (panel d: median 3.8 mg/L; 99th percentile 15.8 mg/L).
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 43
a)
b)
c)
d)
Figure 5-1: Surface SSC time series at near-shore optics survey site 11 (Whanganui) with derived statistical parameters. a) Natural SSC from a simulation with only natural sediments. b) Natural SSC from a simulation with natural and mining derived sediments (suspended source A). c) As panel b but mining-derived SSC. Note the different vertical range in this panel. d) As panels b and c but natural plus mining-derived SSC.
44 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
d)
Figure 5-2: Surface SSC time series at a site 8 km from mining source A. a) Natural SSC from a simulation with only natural sediments. b) Natural SSC from a simulation with natural and mining derived sediments (suspended source A). c) As b but mining-derived SSC. d) As b and c but natural plus mining-derived SSC.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 45
5.2 Suspended sediment source A
5.2.1 SSC The figures in this section show maps of median and 99th percentile SSC in the model layers next to the surface (labelled “surface” or “near-surface” SSC) and the bottom (labelled “bottom” or “near-bottom” SSC) for the suspended source at location A. (For a more precise description of these terms, Appendix A describes the vertical grid layout of the model and shows how the thickness of the model layers depends on depth.) Figure 5-3 shows the median, near-surface SSC. Panel a shows natural SSC, from the simulation with natural sediments only; panels b and c both show results from a simulation with natural and mining- derived sediments, the former showing mining-derived sediments only and the latter the sum of mining-derived and natural sediments. The methods of calculating these statistics are as described for the time series plots in Figure 5-1 and Figure 5-2, but applied to every grid cell in the model. The colour scale runs from 0.3 mg/L (below which is shaded white) to 300 mg/L (above which is shaded yellow). The same colour scale is used for all plots of near-surface SSC in this section.
The median natural SSC (panel a) is similar to, but generally lower than, the mean natural SSC shown in Figure 4-1. There is a strip of elevated values adjacent to the coast, and this strip is wider over the Patea Shoals than it is further to the southeast. The highest values are in excess of 20 mg/L (magenta) and occur near the mouths of the Manawatu and Whanganui Rivers and also in a strip along the coast near Patea. At the inner end of the project area the median natural SSC is close to 1 mg/l (the blue-purple transition). The map of median mining-derived SSC (panel b) indicates that the sediment plume frequently extends to the southeast from the source location towards the coast at Whanganui. The highest values occur around the source and there is an region extending some 5 km from the source with values between 2.5 and 5 mg/L (light green). There is an extensive area with median SSC above 0.6 mg/L (light blue) and a secondary area also above 0.6 mg/L near the coast from Whanganui to Patea. However near the shore the natural concentrations (panel a) exceed the mining-derived concentrations (panel b) substantially. Comparing the maps of median natural plus mining-derived SSC (panel c) and natural SSC only (panel a) indicates that a clear signature of the mining-derived plume is only seen at more than about 8-10 km from the shore, particularly around the source location where the green-shaded area (> 2.5 mg/L) extends considerably further offshore in panel c than in panel a.
46 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
Figure 5-3: Median near-surface concentration of suspended sediment from mining source A. a) Natural SSC; b) mining-derived SSC; c) natural plus mining-derived SSC. Black lines indicate the project area and the 22.2 km territorial limit. An open circle in panels b and c indicates the source location.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 47
Figure 5-4 is similar to Figure 5-3 but shows the 99th percentile near-surface SSC rather than the median. For natural sediments (panel a) this varies from a minimum of less than 2.5 mg/L (purple) offshore from Manawatu and Horowhenua up to approximately 100 mg/L near the coast, with maxima around the Manawatu and Whanganui Rivers and a couple of grid cells over 300 mg/L (yellow) near the Whanganui River mouth. At the inner end of the project area the 99th percentile natural SSC is between 5 and 10 mg/L (dark green).
The mining-derived SSC (panel b) shows an extensive plume with the 99th percentile above 10 mg/L (dark blue) up to 20 km southeast of the source and values over 20–40 mg/L (magenta) within a few kilometres of the source. Values at the coast southward from Patea almost to the Manawatu River are between 2.5 and 5 mg/L (light green), but as with the median these are well below the natural 99th percentile values next to the coast.
A comparison of the natural plus mining derived SSCs (panel c) with the natural SSCs (panel a) shows only subtle differences except in the vicinity of the source, where the area shaded dark blue (10–20 mg/L) extends further offshore in panel c.
Figure 5-5 shows the median near-bottom SSC. Because SSC typically increases with depth, a different colour scale is used for near-bottom concentrations, from 1 mg/L (below which is shaded white) to 1000 mg/L (above which is shaded yellow). The median, near-bottom natural SSC (panel a) is less than 1 mg/L offshore and increases to around 200 mg/L (red) at the coast. The mining-derived plume (panel b) has a similar shape to the surface plume (Figure 5-3b). Bearing in mind the factor of 3 between the colour scales in the two figures, it is nevertheless apparent that the median mining-derived SSC is not as bottom-intensified as the median natural SSC. (This is also seen in plots of vertical profiles, eg. Figure 5-10 below.) The relative lack of bottom-intensification occurs because the plume of mining- derived sediments is dominated by relatively fine particles, whereas the natural suspended sediments are a mixture of coarse and fine particles, with the former dominating near the bottom and the latter dominating near the surface.
Comparing natural plus mining-derived SSC (panel c) and the natural sediment (panel a) again shows subtle differences except near the source and on the outer Patea Shoals, with the SSC near the source doubled, from ~ 10 mg/L to ~20 mg/L.
Figure 5-6 shows the 99th percentile near-bottom SSC. For the natural SSC (panel a) this statistic is 4–8 mg/L (purple) offshore and > 500 mg/L (dark purple to yellow) along the coast. For the mining-derived SSC (panel b) the highest values (~ 100 mg/L) occur within a few kilometres of the source and there is a shore-parallel strip some 80 km long in which they exceed 8 mg/L (light green), with a secondary maximum at the coast near Whanganui, between Patea and Wainui Beach, and along the Manawatu coast. As with the near-bottom median graphs differences between the natural-only and natural plus mining-derived maps (panels a and c) are visible around the source and on outer Patea Shoals but not elsewhere.
48 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
Figure 5-4: 99th percentile near-surface concentration of suspended sediment from mining source A. a) Natural SSC; b) mining-derived SSC; c) natural plus mining-derived SSC.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 49
a)
b)
c)
Figure 5-5: Median near-bottom concentration of suspended sediment from mining source A. a) Natural SSC; b) mining-derived SSC; c) natural plus mining-derived SSC.
50 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
Figure 5-6: 99th percentile near-bottom concentration of suspended sediment from mining source A. a) Natural SSC; b) mining-derived SSC; c) natural plus mining-derived SSC.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 51
5.2.2 Deposition The figures in this section show rates of sediment deposition associated with the suspended sediment source at location A. Deposition is characterised by two statistics, the maximum 5- day deposition (i.e. the maximum increment in thickness over any 5-day interval) and the maximum 365-day deposition. Both are evaluated over the last two years of the simulation.
Figure 5-7 shows the maximum 5-day deposition. The organisation of this figure is similar to that established for the SSC figures in the previous section, Figure 5-3 to Figure 5-6, i.e. panel a is for natural sediment in a simulation with only natural sediments, panel b is for mining-derived sediments in a simulation with both natural and mining-derived sediment, and panel c is for both natural and mining-derived sediments in the same simulation as panel b. Panels a and c relate to changes in the total thickness of the sediment bed, expressed in millimetres. For panel b, we are considering only a subset of the sediment classes in the bed: in this case an equivalent thickness is calculated by taking the total mass of these sediment classes in kg/m2, divided by the in situ bulk density. The latter is equal to the grain density (3100 kg/m3 for the mining-derived sediments) divided by (1-porosity). With the assumed porosity of 0.4, the bulk density is 1860 kg/m3.
Natural 5-day deposition rates (panel a) vary widely from < 1 mm (white) offshore to ~ 20 mm (red) at the coast near the Whanganui River. At the inner end of the mining area they are between 0.4 and 2 mm (light green to purple). The deposition footprint of the mining sediments (panel b) is a narrow strip, some 100 km long, that approximately follows the 22.2 km territorial boundary, with maximum 5-day increments of between 0.2 and 2 mm. The deposition pattern of natural plus mining sediments (panel c) is similar to the pattern for natural sediments only, but modest increases can be seen relative to panel a along the territorial boundary line.
Figure 5-8 shows the maximum 365-day deposition. For the natural sediments (panel a) there is an area shaded white on and south of the Patea Shoals. Over most of this area the maximum 365-day deposition is negative (which cannot be represented on a logarithmic scale). The negative values occur because the seabed in this area undergoes slow but progressive erosion over the course of the model run. This behaviour limits the model’s ability to simulate long term sediment deposition, as it is not possible to run the model to equilibrium on a time scale of a year or two.
For the mining-derived sediments (panel b) the 365-day deposition shows the same pattern as the 5-day deposition, with somewhat larger values, up to a few millimetres.
Panel c (natural plus mining-derived) is similar to panel a (natural only), but panel c shows larger values around the source location and also in an area 10 km from the coast, about mid-way between Foxton and Whanganui, where there appears to be a deposition centre for the mining-derived sediment.
52 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
Figure 5-7: Maximum 5-day increment in sediment bed thickness for suspended sediment from mining source A. a) Natural sediment; b) mining-derived sediment; c) natural plus mining-derived sediment.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 53
a)
b)
c)
Figure 5-8: Maximum 365-day increment in sediment bed thickness for suspended sediment from mining source A. a) Natural sediment; b) mining-derived sediment; c) natural plus mining- derived sediment.
54 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
5.2.3 SSC vertical profiles This section presents figures showing vertical profiles of the median and 99th percentile SSC at three locations for suspended source A. As in Sections 5.2.1 and 5.2.2, there is a comparison between the natural sediments (from a simulation with natural sediments only), the mining-derived sediments and the combination of natural and mining-derived sediments.
The same two sites are considered as in Section 5.1: the near-shore optics survey site 11 near Whanganui and a site 18 km from the coast and 8 km from mining source A, frequently in the path of the mining plume. For the former site (Figure 5-9) the median and 99th percentiles of the mining-derived sediments are much less than the natural values at all depths, so the mining-derived sediment makes a negligible contribution. For the latter site (Figure 5-10) the mining-derived median and percentile are larger than the natural SSC, except near the bottom.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 55
Figure 5-9: Vertical profiles of SSC statistics for near-shore optics survey site 11 (Whanganui). Median (left) and 99th percentile (right) of natural SSC, mining-derived SSC and total SSC for a release at Source A.
Figure 5-10:Vertical profiles of SSC statistics for at a site 8 km from mining source A. Median (left) and 99th percentile (right) of natural SSC, mining-derived SSC and total SSC for a release at Source A.
56 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
5.3 Suspended sediment source B
5.3.1 SSC The four figures in this section (Figure 5-11 to Figure 5-14) show maps of median and 99th percentile SSC for the suspended source at location B. Panel a (natural sediments only) in each case repeats panel a in the corresponding figure for source A (Figure 5-3 to Figure 5- 6). Panel b (mining-derived sediments) in each case shows a plume that is further offshore and with slower concentrations than for source A. However it is noticeable that differences between panel c (natural plus mining-derived sediments) and panel a (natural sediments) tend to be somewhat more noticeable for source B than source A, because source B is further offshore, in an area that has generally lower natural concentrations.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 57
a)
b)
c)
Figure 5-11:Median near-surface concentration of suspended sediment from mining source B. a) Natural SSC; b) mining-derived SSC; c) natural plus mining-derived SSC.
58 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
Figure 5-12:99th percentile near-surface concentration of suspended sediment from mining source B. a) Natural SSC; b) mining-derived SSC; c) natural plus mining-derived SSC.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 59
a)
b)
c)
Figure 5-13:Median near-bottom concentration of suspended sediment from mining source B. a) Natural SSC; b) mining-derived SSC; c) natural plus mining-derived SSC.
60 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
Figure 5-14:99th percentile near-bottom concentration of suspended sediment from mining source B. a) Natural SSC; b) mining-derived SSC; c) natural plus mining-derived SSC.
5.3.2 Deposition The figures in this section (Figure 5-15 and Figure 5-16) show rates of sediment deposition associated with the suspended sediment source at location B. As with the SSC figures, panel a (natural sediments) repeats panel a in the corresponding figures for source location A (Figure 5-7 and Figure 5-8). The deposition footprint for the mining-derived sediments (panel b) sits 5–10 km further offshore than it does for source A. As with source A, the differences between panel c (natural plus mining-derived sediments) and panel a (natural sediments only) are small. They are a little more marked with source B than with source A, because source B is further offshore in a region where natural deposition rates are somewhat lower.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 61
a)
b)
c)
Figure 5-15:Maximum 5-day increment in sediment bed thickness for suspended sediment from mining source B. a) Natural sediment; b) mining-derived sediment; c) natural plus mining-derived sediment.
62 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
Figure 5-16:Maximum 365-day increment in sediment bed thickness for suspended sediment from mining source B. a) Natural sediment; b) mining-derived sediment; c) natural plus mining- derived sediment.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 63
5.4 Suspended sediment source A: recovery simulation As explained in Section 3.2.2 the “recovery” simulation is designed to investigate the behaviour of the sediments released by suspended source A once the release has stopped. Figure 5-17 shows the thickness of the mining-derived sediment at the time the release stops (after 800 days of operation), and then at 365 and 730 days after that.The spatial pattern is very similar to that seen in the plots of the 5-day and 365-day increments in thickness (Figure 5-7b and Figure 5-8b) The thickness is initially up to 10 mm near the source and around 2– 3 mm in a secondary maximum between the Whanganui and Manawatu Rivers. Over the following two years the mound of deposited sediment slowly deflates and moves south- eastward.
Figure 5-18 shows two plots of the 99th percentile of near-bottom SSC from the recovery simulation, evaluated over 0–365 days and 365–700 days after initialisation, respectively. There is a strip of elevated values coinciding with the area of deposited sediment evident in Figure 5-17 and between the two consecutive years there is a substantial drop in concentration. A couple of points should be noted:
. The 99th percentile may not be a good statistical parameter to use in the presence of a trend, as it tends to respond to the larger values occurring at the beginning of each period.
. The maximum in Figure 5-18a and b maybe an artefact of the initialisation and/or an issue with the model boundaries.
64 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
c)
Figure 5-17:Thickness of mining-derived sediment for the recovery simulation. a) Snapshot at the time the release stops; b) Snapshot 365 days after the release stops; c) Snapshot 730 days after the release stops.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 65
a)
b)
Figure 5-18:99th percentile of near-bottom SSC for the recovery simulation. a) For the period 0–365 days after the release stops; b)For the period 365–730 days after the release stops.
66 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
5.5 Patch source The patch source (Section 3.2.3) is a 3 × 2 km rectangular area representing one year’s worth of de-ored sand deposited on the sea floor at mining source A. The patch is put in place at 2200 days and the simulation continues for another 800 days (this being the length of simulation available).
Figure 5-19 shows the thickness of patch sediment immediately after placement, then after another 730 days. A black rectangle indicates the patch outline. The area inside the patch is yellow, indicating a thickness > 100 mm; in fact it is approximately 1 m, this being the total bed thickness in these simulations. (The particular value does not matter, as the sediment bed is never eroded to a significant fraction of this depth.) Initially there is no sediment outside the patch. After 730 days, patch material is found on the surrounding seabed at thicknesses > 0.1 mm up to 7 km from the patch boundary, with a bias to the southeast, and the maximum thickness outside the patch boundary is 21 mm.
Figure 5-20 shows the 99th percentile of the near-bottom concentration of patch sediments for 0–365 and 365–730 days after patch placement. In the former period, the highest values are found over the patch and are between 70 and 80 mg/L. In the latter period these have reduced somewhat, either because fine material has been stripped from the top of the sediment bed in the patch or because the patch has been overtopped to some extent by seabed sediments from outside. In all cases the area of significant (> 1 mg/L) SSC is confined to within a few kilometres around the patch boundary, no further than the area to which the patch deposits have spread (Figure 5-19).
Surface SSCs from the patch source are vanishingly small and have not been plotted.
The plume generated by erosion of the patch is much less extensive than the plume associated with the suspended source (eg. Figure 5-6b). The highest near-bottom SSCs from the patch source, at around 80 mg/L, are comparable with those resulting from resuspension of natural sediments in the model (eg. Figure 5-6a). This is to be expected, as the patch is similar in nature to the surrounding seabed.
It is probable that this simulation is underestimating the rate at which the patch sediment is eroded and transported, because the model’s formulation of active layer thickness does not take account of the complex bedforms in the area (Section 1.6) and we believe that the model underestimates the rate at which sand is suspended at this site, though it overestimates it at other sites (Section 4.3). However, even allowing for some underestimation by the model, there is no indication that erosion from the patch of de-ored sand will produce plumes that are comparable to those produced by the suspended source. This is because most of the fine material in the tailings discharge has been assumed (perhaps conservatively) to have been released via the suspended source and, of the 38– 90 µm material that is incorporated into the patch, most is at depths of several metres and well out of the reach of surface erosion for a long period.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 67
a)
b)
Figure 5-19:Thickness of sediment from the patch source. a) Snapshot immediately after initialisation. b) Snapshot after 730 days. A thick, black rectangle indicates the patch outline.
68 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
a)
b)
Figure 5-20:99th percentile of near-bottom patch SSC. a) 0–365 days after initialisation. b) 365– 730 days after initialisation.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 69
5.6 Freshwater source Figure 5-21 shows the median and 99th percentile of the surface concentration from the freshwater source (Section 3.2.4), expressed as a percentage. They are low, so that the effect of the freshwater in depressing salinity will not be significant. To estimate the concentration of any dissolved substances in the freshwater, the values in Figure 5-21 can be scaled once the concentration of that substance in the freshwater stream is known.
a)
b)
Figure 5-21:Statistics of surface freshwater concentration from the freshwater source. a) median. b) 99th percentile.
70 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
6 Acknowledgements Several NIWA colleagues provided constructive comments on this report and the work it describes: Alan Orpin, Joanne O’Callaghan, Alison MacDiarmid, Graham Rickard, Iain MacDonald.
ADCP velocity data were provided by Iain MacDonald and SWAN wave output was provided by Richard Gorman.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 71
7 Glossary of abbreviations and terms ADCP Acoustic Doppler current profiler, an instrument for measuring velocity profiles
bathymetry The process of measuring and analysing seafloor depth. A bathymetric data set is often informally called a bathymetry.
clay 1. Sediment with grain size < 2 μm;
2. An adjective describing hydrous aluminium phyllosilicate minerals forming flat, hexagonal sheets.
detiding Estimating the sub-tidal component of a time series. There are various methods, but the most commonly used one is applying a low-pass temporal filter.
mid-depth Used in this report to indicate a depth halfway between the surface and the bottom.
sand Sediment with grain sizes between 63 μm and 2 mm.
silt Sediment with grain sizes between 2 and 63 μm.
SSC Suspended sediment concentration: the mass of suspended sediment per volume of water, typically expressed in milligrams per litre (mg/L).
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 72
8 References Ahmad, M.F.; Dong, P.; Mamat, M.; Wan Nik, W.B.; Mohd, M.H. (2011). The critical shear stresses for sand and mud mixture. Applied Mathematical Sciences 5(2): 53–71.
Ariathurai, C.R., Arulanandan, K. (1978) Erosional rates of cohesive soils. Journal of the Hydraulics Division, 104(2): 279–283.
Beaumont, J., Anderson, T.J., Macdiarmid, A. (2013) Benthic flora and fauna of the South Taranaki Bight. NIWA Client Report, WLG2012-55.
Blott, S.J.; Pye, K. (2001). GRADISTAT: a grain size distribution and statistics package for the analysis of unconsolidated sediments. Earth Surface Processes and Landforms 26(11): 1237–1248.
Booij, N.; Ris, R.C.; Holthuijsen, L.H. (1999). A third-generation wave model for coastal regions 1. Model description and validation. Journal of Geophysical Research 104(C4): 7649–7666.
Carter, L., Garlick, R.D., Sutton, P.J.H., Chiswell, S.M., Oien, N.A., Stanton, B.R. (1998) Ocean circulation New Zealand. NIWA Chart Miscellaneous Series No. 76.
Dean, R.G.; Dalrymple, R.A. (1991). Water Wave Mechanics for Engineers and Scientists. ISBN 981-02-0420-5. World Scientific Publishing Co Pte. Ltd, 352 pp.
Folk, R.L. (1954). The Distinction between Grain Size and Mineral Composition in Sedimentary-Rock Nomenclature. The Journal of Geology 62(4): 344–359.
Gall, M., Pinkerton, M.H., Hadfield, M.G. (2013) Optical effects of an iron-sand mining sediment plume in the South Taranaki Bight region. NIWA Client Report, WLG2013-45, 71 pp.
Green, M.O., Black, K.P. (1999) Suspended-sediment reference concentration under waves: field observations and critical analysis of two predictive models. Coastal Engineering, 38(3): 115–141. doi:10.1016/S0378-3839(99)00044-7
Hadfield, M.G. (2011). Sediment Plume Modelling for South Taranaki Bight Ironsand Mining: Phase 2 Progress Draft Report. NIWA Client Report WLG2011-37, 39 pp.
Hadfield, M.G. (2013). Sediment Plume Modelling for South Taranaki Bight Ironsand Mining: Phase 3 Interim Report. NIWA Client Report WLG2012-51.
Hadfield, M.G.; Zeldis, J.R. (2012). Freshwater dilution and transport in Canterbury Bight. NIWA Client Report WLG2011-54: 39 pp.
Haidvogel, D.B.; Arango, H.; Budgell, W.P.; Cornuelle, B.D.; Curchitser, E.; Di Lorenzo, E.; Fennel, K.; Geyer, W.R.; Hermann, A.J.; Lanerolle, L.; Levin, J.; McWilliams, J.C.; Miller, A.J.; Moore, A.M.; Powell, T.M.; Shchepetkin, A.F.; Sherwood, C.R.; Signell, R.P.; Warner, J.C.; Wilkin, J. (2008). Ocean forecasting
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 73
in terrain-following coordinates: Formulation and skill assessment of the Regional Ocean Modeling System. Journal of Computational Physics 227(7): 3595–3624.
Hicks, D.M., Shankar, U., McKerchar, A.I., Basher, L., Lynn, I., Page, M., Jessen, M. (2011) Suspended Sediment Yields from New Zealand Rivers. Journal of Hydrology (New Zealand), 50(1): 81–142.
John, S.A.; Challinor, S.L.; Simpson, M.; Burt, T.N.; Spearman, J. (2000). Scoping the assessment of sediment plumes from dredging. CIRIA C547. Construction Industry Research and Information Association, London
Kalnay, E.; Kanamitsu, M.; Kistler, R.; Collins, W.; Deaven, D.; Gandin, L.; Iredell, M.; Saha, S.; White, G.; Woollen, J.; Zhu, Y.; Leetmaa, A.; Reynolds, R.W.; Chelliah, M.; Ebisuzaki, W.; Higgins, W.; Janowiak, J.; Mo, K.C.; Ropelewski, C.; Wang, J.; Jenne, R.; Joseph, D. (1996). The NCEP/NCAR 40-Year Reanalysis Project. Bulletin of the American Meteorological Society 77(3): 437–471. doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2
MacDonald, I.T., Budd, R., Bremner, D., Edhouse, S. (2012). South Taranaki Bight Iron Sand Mining: Oceanographic measurements data report. NIWA Client Report, HAM2012-147: 109 pp.
MacDonald, I.T., Gall, M., Bremner, D. (2013). Near-shore optical water quality in the South Taranaki Bight. NIWA Client Report, HAM2013-040: 85 pp.
MTI Holland (2013a). Assessment of sediment deposition and re-suspension behaviour of tailings. Technical Report DZ 57.
MTI Holland (2013b). Assessment of sediment deposition and re-suspension behaviour of tailings. Phase 2: influence of surface waves. Technical Report DZ 58.
Nodder, S.D.; Edwards, N.A.; Burrows, M.W. (1992). Patea Sediments. NZ Oceanographic Institute Chart Coastal Series.
Orpin, A.R. (2011). Schedule N: Sediment characterisation analysis: particle size of samples from RC core STH004. NIWA Client Report WLG2011-44, 22 pp.
Orpin, A.R. (2012). Schedule N, Part 2: Sediment characterisation analysis: particle size of samples from cores STH010RC and STH012RC. NIWA Client Report WLG2012-23. 23 pp.
Pinkerton, M.H., Schwartz, J.N., Gall, M., Beaumont, J. (2013) Satellite ocean- colour remote sensing of the South Taranaki Bight from 2002 to 2012. NIWA Client Report WLG2013-14. 79 pp.
Reynolds, R.W.; Smith, T.M.; Liu, C.; Chelton, D.B.; Casey, K.S.; Schlax, M.G. (2007). Daily high-resolution-blended analyses for sea surface temperature. Journal of Climate 20(22): 5473–5496. doi:10.1175/2007JCLI1824.1
74 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Rickard, G.J.; Hadfield, M.G.; Roberts, M.J. (2005). Development of a regional ocean model for New Zealand. New Zealand Journal of Marine and Freshwater Research 39: 1171–1191.
Smith, S. D. (1988). Coefficients for sea surface wind stress, heat flux, and wind profiles as a function of wind speed and temperature. Journal of Geophysical Research 93(C12): 15467–15472.
Stanton, B.R., Goring, D.G., Bell, R.G. (2001) Observed and modelled tidal currents in the New Zealand region. New Zealand Journal of Marine and Freshwater Research, 35(2): 397–415. 10.1080/00288330.2001.9517010
Soulsby, R. (1997). Dynamics of marine sands. Thomas Telford Publications, London. 249 pp.
Soulsby, R.L., Damgaard, J.S. (2005) Bedload sediment transport in coastal waters. Coastal Engineering, 52(8): 673-689. doi:10.1016/j.coastaleng.2005.04.003
Thompson, R.O.R.Y. (1983) Low-Pass Filters to Suppress Inertial and Tidal Frequencies. Journal of Physical Oceanography, 13(6): 1077-1083. doi:10.1175/1520-0485(1983)013<1077:LPFTSI>2.0.CO;2
van Rijn, L.C. (1984). Sediment Transport, Part II: Suspended Load Transport. Journal of Hydraulic Engineering 110: 1613–1641.
Walters, R.A.; Goring, D.G.; Bell, R.G. (2001). Ocean tides around New Zealand. New Zealand Journal of Marine and Freshwater Research 35(3): 567-579.
Walters, R.A.; Gillibrand, P.A.; Bell, R.G.; Lane, E.M. (2010). A study of tides and currents in Cook Strait, New Zealand. Ocean Dynamics 60(6): 1559-1580.
Warner, J.C.; Sherwood, C.R.; Signell, R.P.; Harris, C.K.; Arango, H.G. (2008). Development of a three-dimensional, regional, coupled wave, current, and sediment-transport model. Computers & Geosciences 34(10): 1284–1306.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 75
Appendix A Model setup details The ROMS vertical grid There are several references throughout this report to surface and bottom (or near-surface and near-bottom) values of quantities like suspended sediment concentration and horizontal velocity. These refer to vertical averages over the model layers adjacent to the surface and the bottom, respectively. The ROMS vertical grid layout is indicated in Figure A-1.
Figure A-1: ROMS vertical grid schematic. The blue lines represent the boundaries between layers in ROMS, using the same number of layers (20) and the same vertical stretching parameters as the inner model. The depth (black line) is and idealised depth profile.
The layers are distributed between the bottom and the surface (and follow the surface as it rises and falls with the tide). The vertical spacing is reduced near the surface and the bottom, with adjustable parameters controlling the degree of this refinement. The thickness of the near-surface and near-bottom layers is therefore dependent on water depth. Sample values are shown in Table A-1.
Table A-1: Inner model layer thicknesses. Thickness of the near-surface and near-bottom layers, along with maximum layer thickness as a function of water depth for the inner model.
Water depth Near-surface Maximum layer Near-bottom (m) layer thickness thickness layer thickness (m) (m) (m) 10 0.27 0.63 0.40 20 0.39 1.36 0.74 30 0.46 2.10 1.07 40 0.53 2.85 1.38 50 0.58 3.61 1.70
76 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Appendix B Composition and properties of sediments Size classes This report uses the same sediment grain size classification (Table B-1) as the GRADISTAT analysis program (Blott and Pye, 2001)
Table B-1: Sediment grain size classification. See Blott and Pye (2001) for a comparison with other classifications.
Size range (μm) Name > 2000 Gravel 1000-2000 Sand Very coarse 500–1000 Coarse 250–500 Medium 125–250 Fine 63–125 Very fine 31–63 Silt Very coarse 16–31 Coarse 8–16 Medium 4–8 Fine 2–4 Very fine < 2 Clay
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 77
Appendix C Model-measurement comparisons
Tidal velocity comparison
Figure C-1: Tidal ellipse comparison (all ADCP deployments). Mid-depth M2 ellipses from measurements (blue) and model (red). The ellipses represent the magnitude and orientation of the tidal velocity variations and the straight line from the origin to the ellipse represents the phase.
78 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Figure C-2: Tidal profile comparison (all ADCP deployments). M2 semi-major amplitude versus height above the bottom from measurements (blue) and model (red).
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 79
Table C-1: Comparison of Mtidal ellipse parameters. Mid-depth M2 tidal ellipse parameters from ADCP measurements and model. Here “ratio” means model value divided by measured value and “diffce” means model value minus measured value.
ADCP Semi-major axis (m/s) Eccentricity Inclination (°T) Phase (°) Site/Deployment Meas. Model Ratio Meas. Model Diffce Meas. Model Diffce Meas. Model Diffce Site 5 Deployment 1 0.17 0.16 0.94 0.02 -0.02 -0.04 113.5 120.9 7.4 210.3 211.1 0.8 Site 5 Deployment 2 0.17 0.17 0.98 0.02 0.00 -0.02 112.5 120.9 8.4 205.8 207.8 2.0 Site 6 Deployment 1 0.30 0.29 0.99 0.12 0.08 -0.04 129.1 132.4 3.3 198.6 196.2 -2.4 Site 6 Deployment 2 0.29 0.30 1.01 0.12 0.10 -0.01 130.4 132.7 2.3 197.3 197.3 0.0 Site 7 Deployment 1 0.30 0.30 0.98 0.16 0.15 -0.01 128.9 134.0 5.1 201.1 199.3 -1.8 Site 7 Deployment 2 0.30 0.30 1.02 0.18 0.18 -0.01 130.8 133.5 2.7 199.2 200.3 1.0 Site 8 Deployment 3 0.33 0.33 1.00 0.18 0.17 -0.01 133.4 132.2 -1.1 206.0 205.0 -1.0 Site 10 Deployment 3 0.27 0.28 1.02 0.17 0.15 -0.02 130.2 137.6 7.3 203.5 201.2 -2.2
80 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Sub-tidal velocity comparison
Figure C-3: Sub-tidal variance ellipse comparison (all ADCP deployments). Scatter plots and variance ellipses of ADCP (left-hand panel) and model (right-hand panel) sub-tidal velocities halfway between the surface and bottom.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 81
Figure C-4: Sub-tidal along-axis velocity comparison (all ADCP deployments). Time series of ADCP (blue) and model (red) sub-tidal velocity components along major axis of the variance ellipse at a depth halfway between the surface and bottom.
82 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Figure C-5: Sub-tidal across-axis velocity comparison (all ADCP deployments). Time series of ADCP (blue) and model (red) sub-tidal velocity components perpendicular to the major axis of the variance ellipse at a depth halfway between the surface and bottom.
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 83
Table C-2: Mid-depth sub-tidal velocity comparison. Mid-depth sub-tidal mean and variance ellipse parameters from ADCP measurements and model, and temporal correlations between measured and modelled time series. Here “ratio” means model value divided by measured value and “diffce” means model value minus measured value
ADCP Site & Mean magnitude Mean direction Semi-major axis Eccentricity Inclination Correlation Deployment (m/s) (°T) (m/s) (°T) Meas. Model Diffce Meas. Model Diffce Meas. Model Ratio Meas. Model Diffce Meas. Model Diffce Along- Across- axis axis Site 5 Deployment 1 0.08 0.05 -0.03 97.5 103.2 5.7 0.11 0.06 0.60 0.09 0.15 0.06 95.9 100.5 4.5 0.90 0.38 Site 5 Deployment 2 0.04 0.04 0.00 93.5 103.3 9.8 0.13 0.09 0.68 0.1.0 0.17 0.07 97.0 103.8 6.9 0.93 -0.06 Site 6 Deployment 1 0.05 0.03 -0.02 105.1 104.3 -0.8 0.08 0.05 0.66 0.15 0.19 0.04 106.8 105.0 -1.8 0.88 0.56 Site 6 Deployment 2 0.01 0.02 0.01 108.0 97.0 -11.0 0.11 0.08 0.77 0.16 0.20 0.04 110.5 105.8 -4.7 0.89 0.55 Site 7 Deployment 1 0.03 0.03 0.00 69.1 67.1 -2.0 0.07 0.06 0.88 0.31 0.22 -0.09 95.0 94.3 -0.7 0.87 0.69 Site 7 Deployment 2 0.01 0.01 0.00 35.7 26.7 -8.9 0.10 0.10 1.00 0.22 0.24 0.01 104.0 102.6 -1.4 0.85 0.55 Site 8 Deployment 3 0.06 0.09 0.04 111.2 101.7 -9.5 0.11 0.1 0.93 0.26 0.22 -0.04 98.5 91.7 -6.8 0.92 0.84 Site 10 Deployment 3 0.02 0.03 0.01 111.8 97.5 -14.3 0.07 0.07 0.91 0.42 0.39 -0.04 98.7 96.8 -1.9 0.86 0.72
84 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling
Table C-3: Near-surface subtidal velocity comparison. As Table C-2 but for an ADCP level 5 m below the surface.
ADCP Site & Mean magnitude Mean direction Semi-major axis Eccentricity Inclination Correlation Deployment (m/s) (°T) (m/s) (°T) Meas. Model Diffce Meas. Model Diffce Meas. Model Ratio Meas. Model Diffce Meas. Model Diffce Along- Across- axis axis Site 5 Deployment 1 0.12 0.1 -0.02 102.9 100.3 -2.6 0.12 0.09 0.72 0.13 0.16 0.03 96.3 100.0 3.6 0.92 0.43 Site 5 Deployment 2 0.07 0.08 0.01 110.2 107.6 -2.6 0.17 0.13 0.77 0.11 0.15 0.04 90.4 95.1 4.7 0.95 0.72 Site 6 Deployment 1 0.05 0.04 -0.01 111.1 106.3 -4.9 0.09 0.07 0.77 0.17 0.2 0.03 105.6 106.1 0.5 0.89 0.68 Site 6 Deployment 2 0.02 0.03 0.01 117 111.1 -5.9 0.12 0.10 0.88 0.17 0.25 0.07 110.6 106.4 -4.1 0.88 0.39 Site 7 Deployment 1 0.04 0.05 0.01 77.8 77.7 -0.1 0.08 0.08 1.08 0.35 0.25 -0.10 94.0 97.7 3.7 0.86 0.76 Site 7 Deployment 2 0.01 0.02 0.00 48.4 59.9 11.5 0.11 0.13 1.16 0.29 0.21 -0.08 104.3 103.4 -1.0 0.87 0.58 Site 8 Deployment 3 0.05 0.11 0.06 98.6 96.6 -2.0 0.12 0.13 1.10 0.29 0.28 -0.01 98.4 95.2 -3.1 0.94 0.89 Site 10 Deployment 3 0.03 0.05 0.02 106.7 83.3 -23.4 0.08 0.09 1.12 0.39 0.41 0.02 96.9 99.0 2.1 0.64 0.66
South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling 85
Table C-4: Near-bottom subtidal velocity comparison. As Table C-2 but for the lowest ADCP level.
ADCP Site & Mean magnitude Mean direction Semi-major axis Eccentricity Inclination Correlation Deployment (m/s) (°T) (m/s) (°T) Meas. Model Diffce Meas. Model Diffce Meas. Model Ratio Meas. Model Diffce Meas. Model Diffce Along- Across- axis axis Site 5 Deployment 1 0.05 0.02 -0.02 88.3 97.2 8.9 0.07 0.04 0.58 0.16 0.27 0.11 104.4 116.4 12.0 0.83 0.61 Site 5 Deployment 2 0.02 0.02 0.00 26.7 22.5 -4.2 0.08 0.05 0.61 0.17 0.22 0.05 117.5 127.6 10.1 0.89 0.88 Site 6 Deployment 1 0.03 0.02 -0.01 107.1 105.6 -1.5 0.06 0.04 0.56 0.15 0.28 0.13 108.8 103.9 -5.0 0.85 0.28 Site 6 Deployment 2 0.01 0.01 0.00 278.4 87.1 168.6 0.08 0.05 0.68 0.18 0.33 0.15 112.9 103 -9.9 0.90 0.42 Site 7 Deployment 1 0.02 0.02 0.00 69.0 62.1 -7.0 0.05 0.04 0.83 0.33 0.24 -0.08 99.2 93.1 -6.1 0.90 0.29 Site 7 Deployment 2 0.01 0.01 0.00 298.0 22.2 84.2 0.07 0.07 0.96 0.26 0.3 0.04 108.4 99.9 -8.6 0.88 0.33 Site 8 Deployment 3 0.04 0.07 0.03 126.2 108.9 -17.3 0.08 0.08 0.96 0.32 0.24 -0.08 102.2 92.0 -10.2 0.93 0.57 Site 10 Deployment 3 0.02 0.02 0.00 164.5 145.9 -18.6 0.06 0.04 0.74 0.5 0.52 0.02 123.8 101.3 -22.5 0.81 0.39
86 South Taranaki Bight Iron Sand Extraction Sediment Plume Modelling